Good morning. Welcome back to this morning session. I'm Vivek Arya from BofA Semiconductor Semicap Equipment Research Team. I'm really delighted to have Jim Anderson, President and CEO of Lattice Semiconductor, join us this morning. We'll go through my Q&A, but if you'd like to bring anything up in between, please feel free to raise your hands. Very warm welcome to you, Jim.
Thank you, Vivek.
Really appreciate you joining us.
I'm glad to be here. Thanks for giving us some time. Appreciate it.
Yep, my pleasure. I was wondering if you could just maybe start with the state of the union, right?
Sure.
Again, in terms of just the broad demand trends that you're seeing in your end markets versus what you thought at the start of the year?
We've had a pretty good start to the year, right? First of all, in our most recent quarter, Q1, we felt pretty good about Q1. That's our, let's see, 12th consecutive quarter of sequential revenue growth. We grew 22% year-over-year. We saw growth from both of our big end markets of communications and computing, and industrial and automotive, both grew on a year-over-year basis. Industrial auto grew even stronger than the other segment. So we saw, you know, really good, healthy Q1, so good start to the year. If you look at Q2, the current quarter, in terms of our guidance, if you take the midpoint, we guided sequentially up, and we certainly have a lot of new product cycles that we're going through right now. When we look further out over the long term, you know, we just did our Investor Day, our Analyst Day.
Right.
About a month ago, you know, over the long term, we continue to see, you know, great strong growth in our core markets, and that really being driven by all the new product cycles that we have going.
Got it.
Yep.
Makes sense. Now, the one question that, you know, comes up a lot is, you know, how to look at, these different technologies, right, that are available, FPGAs-.
Sure.
-versus microcontrollers. If you could start with the technology side and then, you know, get down to your specific products. Where do FPGAs stop, microcontrollers begin, or I guess, the other way around, for you? How does that shape how you invest your R&D dollars and how you look at your growth objectives?
I don't think it's so much as, here's where FPGAs start, or here's where microcontrollers begin. It's more, you know, there are different silicon tools, right? an FPGA, a microcontroller, a general purpose CPU, a graphics processor, all these are different tools in the toolbox.
Right.
Right? Sometimes FPGAs are the right tool to use, and sometimes other types of silicon are. The reason that FPGAs have been used now for, you know, programmable logic's been around for 40 years. Actually, Lattice turns 40 years old this year, right, in April. The reason that programmable logic has been around for so long and has grown over all those decades is because, first of all, it's incredibly flexible and adaptable, so you can program the silicon to do just about any task you want. You can take your own custom algorithm, whether that's, you know, some sort of operational algorithm or an artificial intelligence algorithm, whatever that is, you can program that algorithm right onto the device, and then, very importantly, you can change that over time.
FPGAs oftentimes give you some level of future-proofing of your system, because let's say, you're in production for three to five years, and you want to tweak the system, you just reprogram the FPGA to incorporate new functionality. FPGAs play just an important part in the overall silicon toolbox. You know, we continue to see really good growth across our four core markets for FPGA technology, and obviously, we're investing significantly in a huge product portfolio expansion.
Got it. One thing you outlined also at the Analyst Day was just the ramp in your Avant mid-range.
Mm-hmm.
product and how you have been able to successfully convert Nexus, right, customers to the product. What's the impact of...
That we can address with our customers. That combination of continuing to build out our small FPGA portfolio, plus adding Avant, that means that right now, Lattice is going through the biggest product portfolio expansion we've ever done in our 40-year history. That's what we're really excited about. That's, by the way, that's what our customers are really excited about, too. If you look at Avant, if you look at, first of all, the target customers, 90% of the target customers are already customers of Lattice today. That's really good because penetrating a new customer is a lot harder than just selling another product line to an existing customer.
Right.
We're selling another product line to existing customers. Software is a big factor in our customers, right? The software that our customers are already using on, for instance, Nexus devices or pre-Nexus devices, they can use that software on Avant as well. They can leverage that same software, so they don't have to go learn a whole bunch of new software.
Right.
We've made the adoption path from moving from a Nexus device to an Avant device very, very smooth. We've made the, as we call it, as frictionless as possible, right? We want it to be really easy for those customers, existing customers to adopt Avant. If you look at the design win funnel for Avant, relative to, for instance, Nexus at the same point in time, it's significantly bigger than Nexus at the same point in time of the launch of the platform. We're pretty excited about the, you know, the rate and pace of adoption. In terms of revenue, I think that was somewhere in your six questions, I think.
Too much caffeine.
Yeah. I know how that feels. We expect Avant revenue a little bit before the end of this year, but it would be a bigger contributor next year, and then it would ramp in future years. The other great thing about Avant is the ASPs are 10-20 times higher than our ASPs today for small FPGAs. That revenue ramp of Avant is totally greenfield. It's totally additive for the company because we, this is a brand-new part of the market with existing customers. Every $1 of Avant revenue is, by definition, an additive, greenfield, new growth area for the company. We believe that as Avant ramps over time, it becomes a bigger percentage of our revenue, it increases our revenue growth rate over time.
Right. You mentioned 10-20 times-
Yeah, higher ASPs.
Higher ASPs. Obviously, who cares about cannibalization at that point, right? If you are able to offset it.
Avant really doesn't cannibalize Nexus or pre-Nexus devices because those products are addressing two different capacity points. If a customer had a particular application that they needed something done, they would either select. If they needed a higher capacity device, they'd select Avant. If they needed a lower capacity device, they'd select Nexus. It's not like they would use an Avant instead of a Nexus, right? There's really no cannibalization that happens.
Got it.
Yeah.
Makes sense. You said because of the software base, it's easier for them to make that?
Yeah, absolutely. We pre-plumbed the software, so even when we were, you know, we've been ramping Nexus now for a number of years. The software that they're using on Nexus, we tried to pre-plumb that for Avant so that when Avant Silicon became available, that software was essentially already ready for Avant adoption.
Got it. Makes sense. How did you come to this strategic decision, Jim, to launch the mid-range product at this time, and what does the competitive landscape look like?
Yeah. The first part of the question's really easy. I wish we could take credit for it. It was actually our customers that told us to go build Avant. It wasn't our idea. In 2019, we were introducing the Nexus platform and the roadmap to our customers, so we were going out to all of our strategic customers and saying, "Nexus, which is for the small FPGA part of the market, hey, here's the new roadmap we just put in place." We got really good feedback from customers on the roadmap, but also the competitiveness of Nexus. Super power efficient, much higher power efficiency than our competition. Some of our big strategic customers said, "Hey, we really like the Nexus products, but why aren't you guys building higher capacity devices?
Like, why don't you just build mid-range FPGAs as well?" In particular, for instance, like industrial and automotive customers would ask us that because they don't use large FPGAs. They mostly just use mid-range and small FPGAs. Their point was, "Hey, introduce a mid-range device, and you can pretty much cover all of our FPGA needs, and then you're kind of an end-to-end supplier for us," right?
Right.
That'd be a good thing." We had enough of these strategic customers ask us that we took a look and decided to start investing. We took that power-efficient architecture and expertise that we had in Nexus and that we built up over decades of experience, we took that power-efficient architecture and then built a purpose-built architecture for the mid-range part of the market. At the end of the day, it was our customer's idea. By the way, we went back to those customers once we decided to invest in 2019, and we said, "Hey, we're gonna go invest in this product that you want, but come along for the ride, right? Be there with us through the development cycle, guide us on the features, capabilities that you need so that when the device is ready, we know that we built something that's exactly what you need, right?
Right.
This really customer-driven innovation is what we would call it, right? The customers, and now you can imagine, you know, a few years later, now they're really happy, right? Because they got exactly. We did what they asked us to do, but we also built the product that they want.
Right.
That shows really good strategic partnership from our part over the long term as well. Anyway, the customer's idea. On the competitive landscape, we feel really good about the competitive landscape. When we launched Avant in December, we showed measured silicon data. This isn't made-up, like, spreadsheet stuff, right?
Right.
Measured data of Avant Silicon versus, comparable devices that our customers would be, you know, selecting from our two traditional competitors, Xilinx and Altera, now part of, you know, Intel and AMD, obviously. We did measured comparisons, and we measured for power efficiency, 2.5 times better power efficiency. Not like... That's not 10% or 20% better, 2.5 times better power efficiency. Power efficiency is a big deal to every customer that we have because power, the total system power envelope, is almost always a primary design constraint, when we bring 2.5 times better power efficiency, that's a big deal. Higher performance, the actual physical device size is up to six times smaller, that's a big, that's a big savings on a system board.
If we can bring the solution that's 2.5 times better power efficiency in size, area, physical area, that's six times smaller, that's a big win for them. It's very, very competitive. We feel very good about how it's competitively positioned in the market.
Got it. Now, one thing that those competitors will say is, "Well, yes, we lost share in the last two years," all right? Which is clear to see.
Yeah.
You know, that was because we didn't allocate enough capacity to it, right?
Right.
As we get, you know, the capacity side sorted out, that.
Yeah
"We will come back and start to regain that share." How would you address that part of the story?
I would say, if that's their story, good luck with that. Yeah. What I would say is, if you think about it strategically from a customer's perspective. Put aside the product, right, for the moment, just strategically, and if you take, like, a large industrial customer, for example, first of all, FPGAs, but this would apply to other customers as well, but FPGAs are an important part of their silicon toolbox, right? They like FPGAs because they can customize them for their particular piece of equipment, because they can adapt and change it over time, et cetera. They have three general purpose suppliers, right? They have Altera, Xilinx, and us.
I think what we've demonstrated over the past years, certainly over the, you know, almost five years I've been part of the company, is, number one, we're willing to invest in the type of products that are important to them and their business, which is small and mid-range FPGAs. We're investing, we're innovating, we're building software for them, so we're naturally strategically aligned to them in terms of the roadmap of products that we're building, right? The second is, if you look at our revenue base, our revenue is naturally aligned to the markets that they're in. Third, yeah, you're right. When times got tough, when supply got tight, they saw that their other two suppliers allocated supply to data center, for example. We didn't. We stuck by our industrial and automotive customers.
We made sure that they were supplied. You know, we weren't perfect. We had our own supply issues, but I think our customers would agree that we did, we did a, you know, a great job relative to, you know, the situation in terms of making sure they were supported, and they remember that. I think our customers are naturally strategically aligned...
Right.
to shift share to Lattice. I think when you add on top of that, the fact that we've actually got a better product, I think, yeah, the customers are really excited about Lattice and growing their relationship with Lattice. You know, I can tell you, five years ago, when I joined the company, we weren't having long-term strategic discussions with our, you know, our big top customers. That just didn't happen. In fact, I would have a hard time actually getting customers to meet with me five years ago. That is totally 180 degrees different now, right? We're having really great, deep, long-term strategic customer or discussions with all of our big strategic customers, which is great.
Where would you put Microchip, Microsemi as part of the competitive landscape? Do you run into them in an industrial-
We do run into them, but it's mostly in defense and military applications.
Micro-
Yeah, and I think that kind of goes to where their heritage, that's the old Actel business.
Right.
That kind of goes back to the heritage of that business. If we're competing in a defense and aerospace application, we would typically run into them. In the more general purpose applications, which is the large majority of our TAM, the typical competitors would, for us, would be Altera and Xilinx.
Got it. Makes sense. next thing, how would you know, contrast FPGAs versus custom ASICs, especially in the comms and data center, market? right, because, you know, we saw, for example, the battle between, say, a Xilinx and a Marvell, right...
Yeah.
go from FPGA to custom ASIC.
Yeah.
I know that's maybe a slightly different kind of workload.
Yeah.
How do you see the custom ASIC competitive situation?
Well, we really don't worry about that. It's very rarely, because we're building mid-range and small FPGAs, it's very rare that we're replaced by an ASIC. It usually doesn't make technical sense or economic sense to do that.
Right.
From a technical perspective, usually when the customer is designing us in, they're designing us in because they want the flexibility to be able to reprogram the FPGA over time, right? They want that future-proof benefit of being able to upgrade or change the feature set to the FPGA over time. We really don't worry about ASIC replacement. It's very rare that we see that. You're right, there are other instances in other app, I would say, more specific applications. Like, I think the one you're referring to is baseband processing and base stations.
That's right.
Yeah. There, and then, that's where, customers will use very large FPGAs in the first generation. These are very large FPGAs that are very, very power hungry. They'll start with that generation, and then kind of get their algorithm figured out, and then they'll put that into an ASIC in the second generation. There are instances with the very large FPGAs, but for mid-range and small, we just don't see a lot of ASIC replacement.
Got it. I have to ask the obligatory AI question in this.
Vivek, I can't believe it took you this long to ask an AI question. Come on, you're not doing your job.
If I bring it up too early, it's not great. If I bring it up too late, it's not good either. I did bring it up.
Oh, an AI question, yeah.
How does it impact Lattice?
Yeah, it's good for us, is the short answer. There's two places that we participate in terms of AI. One is at the edge of the network in AI algorithms. This would be more inference processing that's being done at the edge of the network. The other is in the data center, in servers or more AI-optimized servers as well. If I start with the data center, and this, whether it's general purpose servers or it's the servers that are sort of more optimized for AI processing, maybe have more GPU content, if you look over the last three, four years, actually, we've built up a really good position in servers. In servers, we're doing control, management, and security of the platforms. We're not part of the compute complex.
We are doing that management and security. We're making sure that the platform itself is secure so that when the server platform boots up, that the firmware hasn't been corrupted, that the hardware is secure, et cetera, making sure that it can be remotely managed, et cetera. Our attach rates now are well over 1x. When I started, we had a little bit of a toehold in servers. Our attach rates were pretty low. We've grown that to over 1x, which means, almost every server in the world ships with at least one Lattice piece of silicon, and a lot of them actually ship with multiple pieces of Lattice silicon.
We've been able to expand our attach rate, expand our ASPs, as well as we brought more content, our dollars of content in servers has continued to grow. To the extent that, you know, artificial intelligence, generative AI, whatever, drives demand for compute in general in the data center, we benefit from that. Even if it's more AI-optimized compute, we see that as a benefit as well. In an AI-optimized server, we usually have equal to or greater than content in an AI-optimized server. We see it as a net benefit for us in the data center. The other place that we benefit is at the edge of the network. We're seeing a lot of customers that are trying to add more intelligence, more decision-making capability to their systems.
These could be industrial customers or automotive customers, lots of different types of applications. They're trying to make their edge system more aware of its surrounding, able to adapt. Invariably, what they're trying to do is then they're trying to add some level of artificial intelligence processing. This is usually inference processing. This is not the training of the network, but using the trained neural network to make decisions locally. The reason they want to add that intelligence at the edge is they want that equipment to be more autonomous as well, because they don't want to rely on it absolutely having 100% connection back to the data center at all times. They want that piece of equipment to be intelligent and autonomous, even without a constant signal back to the data center.
Anyway, when they're adding this inference at the edge, they're finding that Lattice FPGAs work really well for that, because what we can do is we can take that inference algorithm, which is a parallel algorithm, we can map that onto a Lattice, a power-efficient Lattice FPGA, and we can get really good performance per watt on a number of different AI applications. That's one of the other places that we're seeing really nice growth and adoption, and where we're benefiting from AI.
Got it. Is it?
Just for the record.
Sure.
We were talking about that long before.
At least-
Yeah, exactly. You know, I think we were talking about that two to three years ago at least.
Got it. Is there a way to quantify how much of your portfolio is impacted?
I heard that's popular now to do that.
Yes
... right? We haven't, we haven't broken out yet a quantification of how much of our revenue is from AI. There's certainly a good portion of it. Maybe sometime, maybe in the future, we'll break that out. We haven't done that to date.
Understood.
Yeah.
More importantly, I think, you know, you've put a lot of emphasis on attaching more software, to your...
Yeah
... product.
Yeah.
I think you, really, you know, mentioned that at the Analyst Day. What percentage of your products now carry software with it? How much of an adder is it, and, what's the goal longer term in terms of the software attach rate?
Yeah, this is a really important question. This is a key part of our strategy, and I still think this is a more underappreciated part of our strategy. Let me first explain the strategy, and then I'll answer your specific questions. When I began about five years ago, we started ramping up the software investment in the company. We continued to actually invest in software at a faster rate than even our hardware or silicon development, even though we're growing both. Our strategy for software is very, very simple. It was, we wanted to develop software that makes it really easy for our customers to adopt Lattice silicon and solutions overall, to make it easy to switch from a competitor's device to our device, or if they've never used an FPGA, to adopt us, Lattice silicon really easily.
What we did is we started building a portfolio of application-specific solution stacks. Think about these as prebuilt software tools, libraries, reference designs, but a prebuilt solution stack that's specific to a particular usage model that was common across many customers. We built out a portfolio of these now, and what customers do is they use that along with our silicon, and it really speeds up time to market. Sometimes we've seen customers get to market three to six months faster, which is a big deal to them, by using our software. We're now measuring the adoption rate at over 50%.
If you look at design wins, these are silicon design wins over the last 12 to 18 months, over half of those silicon design wins now have a software attach, meaning a customer is using one of those five, soon to be six, solution stacks that we have. By the way, the attach rate when I, you know, four or five years ago, was zero. It's gone from zero to 50%, so we're pretty happy with that. The other thing that we measure is for those silicon design wins that have a software attach, we measure the ASP of those design wins versus without a software attach. What we can see is that the ASPs of those design wins are higher.
Those, the ASPs, with software attach are higher, but also because of the, you know, software doesn't have COGS, the gross margin of those design wins are higher, and as they transition into revenue, that's a natural tailwind for our gross margin. We know that the customer is paying us for the value that we're bringing through software, and we can also calculate an ROI for that as well. The most important thing, beyond just making it easier for our customers to adopt our solution, is we also believe it creates long-term stickiness. As customers integrate that into their system-level software or into their system in general. That higher level content that we're providing creates multigenerational stickiness. That, we believe, has a very positive long-term benefit. It's certainly, Vivek, it's certainly part, a key part of our strategy.
Got it. How much of an ASP uplift?
We've only qualitatively said that it's a significant uplift. We haven't quantitatively broken that out.
Got it.
Yes.
Thanks. on gross margins.
Yeah.
I think you set a target of, low 70s.
Yep.
You're there already. So as the software attacks it,
Do you think?
That's a good problem to have.
That's too low? That's too low.
No.
Vivek is complaining.
Pretty, pretty good. What are the kind of the drivers of taking it higher? Is there a limit, or like, what...?
Yeah, no, a good question. Definitely gross margin expansion has been a key part of our focus. We started our current gross margin expansion strategy, we kicked that off in the beginning of 2019, and since we started that strategy, we've expanded gross margins by over 1,300 basis points over the past, you know, four, going on five years. We remain focused on gross margin expansion. I do think there continues to be opportunity for us to continue to optimize gross margin. Some of the software that I just talked about, the benefits from software driving higher ASPs, you know, some just new product cycles in general. We're going through so many new product cycles right now. New product cycles are always an opportunity to reoptimize the value equation with our customers.
We're always focused on, you know, continuous product cost reductions, et cetera. I do think there is opportunity to continue to optimize gross margin. We remain focused on that. I will say that our top priority is to drive revenue growth, right? Priority number one is driving top-line growth as quickly as possible, but we do continue to focus on gross margin as well.
Got it.
Yeah.
You laid out, very strong, 15%-20% kind of growth CAGR that is still well above...
Mm-hmm.
your closest, right, peer group.
Yeah.
One kind of near-term question: How are you thinking about the second half? Because there do appear to be a lot more, you know, macro headwinds, right, and some of the growth rates, as we have heard about, you know, for example, AMD's embedded business, right, slowing down into the backup, et cetera. What is the near-term kind of setup, macro-wise, looking like for you?
Yeah, we, the way we always look at it is, We're certainly affected by end markets just like everybody else. We're not immune to any, if there is any softness in end markets, we're certainly affected by that like everybody else. Where we, certainly where I spend my energy, is making sure that whatever that end market growth is, we're growing faster than the end market, right? That we're either driving share gain, that's helping our drive our growth rate faster than end market, or we're driving content expansion. That's where we're driving TAM expansion. We're getting our silicon design into new applications. I think if you look over the last, you know, three, four years, we've got a good track record of doing that, right? Y ou know, communications and computing, one of our two largest segments, that's grown double digits for four years in a row, right?
Right.
If I look at where did that growth come from, it's, that growth primarily came from content expansion, from share gain. In our industrial and automotive segment, that grew three consecutive years at double-digit growth rates. Again, where did that growth come from? It came from share gain, content expansion, et cetera. When we look forward over the next, you know, three, four years or whatever time frame you want to pick, you know, what we're focused on is continuing to drive where we've seen most of the growth come from, which is, again, share gain, content expansion. That's the part, that's what we're focused on. You know, I think the thing that positions us really well in any cycle, but certainly this current cycle, is, again, we're going through the biggest product portfolio expansion we've ever done in the company's history, right?
Right.
We've got so many new products ramping right now. Each one is adding their own new fresh, new revenue stream. That's just a really good place for the company to be. Nexus, we haven't talked. We've been talking about Avant. We haven't talked much about Nexus. Nexus, we now have five different versions, device families of Nexus, that are ramping into production. The fifth one just entered production in Q1 of this year, right? We just launched the sixth one. That'll go into production next year. We have committed to a seventh one later this year. Nexus, we believe Nexus will continue to ramp for years to come, right? Even our pre-Nexus devices continue to grow as well, right? We're still driving growth on the pre-Nexus devices.
I think with so many different product cycles and being positioned in really good long-term secular growth markets, the markets that the company should be in, I think we're really well positioned over the long term.
Terrific. Thank you so much, Jim.
Yeah.
Really appreciate your time.
Vivek.
Thanks for the insights.
Thanks for the time.
Thank you. Thank you.