All right, good morning, everybody. Thank you all. Day two of the TD Cowen TMT Conference. Thank you all for attending. This is kind of the semiconductor room, so a lot of familiar faces. My name is Matt Ramsay from the Semiconductor Research team, and really, really excited to have Sherri and Esam from Lattice Semiconductor here to have a conversation about their company. It's been a remarkable journey since you guys took over the reins of the company. I remember trying to work like heck to get my initiation report out to make sure the stock didn't move from $6-$8. Well done to the whole team. But it's a really, really interesting time.
Esam, maybe you could give a little bit of an overview of where you guys are in the product transition from the market share that you have taken and continue to take in small FPGA. And the company's going through an exciting entry into sort of the mid-tier of the FPGA market with a couple of different customer sets, but a lot of the same customers, but much, much higher ASPs and a different type of application.
Yeah. Thanks, Matt, for having us. So, today, if you look at our product portfolio, it's the strongest product portfolio the company's had in its history, and the company is 41 years old. As of last month, I think we hit 41. And we started this with our Nexus platform launch, which we did in December of 2019, and this we introduced a new modernized small FPGA platform that was very differentiated from a power efficiency, form factor, performance, reliability. And we launched that product in December of 2019, and since then, we've introduced 7 distinct device families based on that platform, of which six have reached initial production, and the seventh one, which we introduced last year, will have the initial production in the second half of this year, and these are very differentiated products.
If you look at the company since then until now, I think we've demonstrated share gains and good growth based on our small FPGAs, which is basically Nexus platform, but also the pre-Nexus as well has been growing. When we introduced that Nexus platform in 2019, with its power efficiency, a lot of our customers came to us at that time and said: "This is a really good power-efficient architecture. What do you have, and can you build something in mid-range?" And what we decided to do at that time is we told our customers, "We can go build a mid-range FPGA. Nothing technically prevents us from doing that, but would you participate in that journey?" And they said yes. So we had about over 100+ customers that helped us define our Avant platform, which is our mid-range FPGA platform.
What was nice about this platform, not only is it very differentiated, again, from power efficiency, performance, and form factor, but 90% of the targeted customers for mid-range FPGA applications are already existing customers of Lattice. They're familiar with our software tools, and we've built these software solution stacks, which we may talk about later, which help our customers get to market much faster. Well, those software solution stacks were designed to support both our Nexus as well as our Avant platform.
We launched the first mid-range FPGA in December of 2022 with our Avant platform launch, and the first device was our Avant E, which is edge-optimized FPGA, which actually we achieved a really good milestone at the end of 2023, which is initial production of revenue from that Avant E device, and we expect that to continue to ramp through 2024 and continue to ramp even further in 2025. And then last year, in December of 2023, we announced two additional device families based on the Avant platform, our Avant G and our Avant X device. G being our general purpose FPGA for mid-range, X being our advanced connectivity FPGA, which opened up more applications for mid-range for us.
And those we expect to hit a milestone of initial production ramp of revenue, although small, at the end of this year, and that will continue to ramp over the next few years. Very differentiated products. And so when I talked about earlier, our strongest product portfolio in our history, it's not just the silicon hardware, it's also on the software side. And what we've done over the past several years is introduced what we call software solution stacks, and we've had now six of these software solution stacks introduced into the market. And they focus on specific types of markets and applications. So we've got sensAI, which is really focused on artificial intelligence on our FPGAs. We've got security ones called Sentry, et cetera.
We've got mVision, all the way to our most recent announcement, which was Lattice Drive, focusing on applications in the automotive market.
I mean, obviously a lot going on there, and some we'll get into more of the product stuff in a minute, but the elephant in the room, I guess, in the FPGA space is we've gone through a heck of an inventory correction, your competitors, yourselves. Sherri, maybe you could spend a little bit of time on from your vantage as CFO, where we are working our way through that, because what I wanna understand is how to get ourselves in the financial model to a level of stability to where we can now grow from, right?
And so we've had a big boom, and then we've come down a bit on inventory correction, and I just, from your perspective, where are we in that process, and like, what are the KPIs that you're looking at to say, "Okay, now we're out the back end of this," and all the exciting stuff Esam's gonna tell us about can make the P&L work.
Absolutely. So thanks, thanks, Matt. Thanks for having us as well. So we are in a cycle and a semiconductor cyclical correction, not just Lattice, but the semiconductor industry. You know, there have been cycles before, and there will continue to be cycles, so this is just one of those cycles. And what we talked about in our last earnings call is that, you know, we expect the inventory digestion to continue into Q2, and that we expect that to dissipate into the second half of the year. And we said that we expect the second half to be higher than the first half in terms of revenue. Why do we think that?
Well, what number one is the inventory digestion and dissipation throughout the second half of the year that we expect to occur. But the bigger thing is really our Lattice-specific growth drivers, and that's really our products. Esam talked about those products, our Nexus products, seven devices on our Nexus platform, and our Avant platform, where we've launched our E, which we generated a smaller amount of revenue at the end of last year, and then G and X expected at the end of this year. So at the end of this year, we'll contribute, you know, to that second half being higher, and our Nexus devices. We have so many devices that we've launched on our Nexus platform.
Each of those devices is additive to revenue in terms of, adding to, and, and contributing to that second half being higher than the first half. So that, that really gives us the, the conviction. It's, it's the fact that, you know, even though we're in this cyclic downturn, you know, we've continued to innovate, and we will continue to innovate. When you look at our our OpEx, for example, you'll see that, you know, from an R&D perspective, you know, we've increased our R&D spend both sequentially and year over year. And, and why is that? Because we wanna continue to make sure that we're investing for the long term of the company.
And in the near term, while we're in this correction, we wanna be disciplined about our OpEx, in particular on the SG&A front, but we also wanna make sure that we are investing for the long term because that's the long-term growth of the company. And so that's very important to us. And, you know, certainly on a quarterly basis, you might see fluctuations in OpEx, but, you know, we're certainly wanting to continue to invest in the long term of the business. The other thing I'll mention is just from an inventory, you know, digestion, we talk about, you know, correction and inventory digestion.
So when we look at our distributor inventory that's in the channel because that's, you know, part of that digestion and we get that question, as well, we really see that our inventory in the channel is at pre-pandemic levels. You know, there might be some devices that are or OPN numbers that are, you know, slightly lower and some that are slightly higher, but overall, we would see that, you know, say that our inventory in the channel is essentially what it was pre-COVID. But and we've got great visibility in terms of what our distributors have in terms of inventory on hand. They provide that visibility to us every week.
But you know, we have over 10,000 customers, and so it's really you know, we don't have visibility to all 10,000 customers in terms of the inventory they have. But we do get inputs from them. We have very close relationships with our strategic customers, and so that gives us the visibility that we think the second half will also be higher than the first half.
Just to follow up on that, Sherri, do you think we get to a point where you're shipping in line with sort of end sell-through at some point in the second half, maybe the Q3 ? I'm just trying to figure out if you've calibrated that a bit more. You've obviously overshipped for a little bit as the inventory built, and then pretty significantly undershipped as we've been trying to flush some of the inventory through the system. But any sense as to when those lines sort of connect and we're back on square footing?
Yeah. Well, I mean, you know, clearly we're, we're undershipping to consumption, that, that we can see, and that really goes back to, you know, customers and customers consuming the inventory that they have on hand, and so we're undershipping, to that demand. I, I would say I would characterize it just a little bit differently, from, from what you had just said a second ago is, I, I don't view that we overshipped into the channel-
Mm-hmm
... or to our customers. I certainly other companies did, and I think that's pretty clear in terms of the verbiage that they provide in their earnings calls, and you see that in their numbers. But, we were very intentional that we did not wanna do that. We wanted to make sure that, you know, we, you know, our strategic customers got the product that they needed, but we didn't see any benefit to either our customers or ourselves in overshipping anything. I think it's just really the visibility that our end customers have to their demand, and then how quickly they can digest that. And so, you know, to your question, will we ship to consumption in the second half?
We've not given that level of color, but certainly over time, you know, we would expect that to go up, and it goes back to those Lattice-specific growth drivers, our Lattice products. But again, second half, we do expect that to be higher than the first half.
Yeah, got it. Makes sense. Just one last question on sort of setting the base for future growth. Esam, I wanted to get your thoughts on, I mean, a pretty good growth driver for the company over the last three or four years has been what the FPGA attach has been in the data center market.
Mm-hmm
... for a couple of different applications. And you guys have now more than one device attached to basically every server that ships by any OEM, including the hyperscale folks. There's been some a conjecture by a couple of your competitors that they might want to take some of that share, let's say. So it'd be interesting to if you could characterize the visibility that you have in the server space, what the functionality is that your products are delivering, what the visibility you have on multiple server generations, not just AMD, not just Intel, not just Arm-based, but just what that profile looks like, and particularly the visibility of holding serve in those markets as that industry grows.
Yeah, I'd love to do that. So, FPGAs, as you said, Matt, are being widely used in servers today, and in fact, if you go back...
... in 2019, at our first Investor Day, we talked about our server opportunity, and that we're investing in there, and we had an attach rate at that time. We talked about close to 25% that we expected to grow to 80%- close to 80%. Now, if you fast forward today, our attach is well over 1. So which means, as Matt said, if you open up a server out there, you're gonna find, if not one, multiple Lattice devices within, within a server. And so what's driven that? What are we actually doing in the servers? Well, we do control, management, and security-type functions within a server. So if you think about, a motherboard, you've got, a lot of functionality in a motherboard.
We manage all the power management, control management, board management, as well as security-type functions, and that's something that's becoming more and prevalent across all servers that you require that. Servers are becoming more complex. You need that programmability. We're also CPU agnostic, which customers like. So it doesn't matter if you're using an AMD, an Intel, an Arm, or some other CPU, what we do is agnostic to that. So as they have different bill of materials, they can use our FPGA. The programmability comes as a benefit for them. And now what we're seeing also, if you look at server architectures, it's becoming more modularized, meaning that it's no longer just a single motherboard.
You have the motherboard that has the CPU on it, or GPU, or whatever that you want to put on there, but they're breaking that functionality into multiple boards versus a single board, and this is a trend that we're seeing in the, in the industry around servers, and they're modularizing it. The reason why they modularize it, it gives them more optionality to take something out and put something in and not have to redesign that entire motherboard like they used to in the past. So you see some of the control and security boards being separate now, and that way, they can remove a CPU, put another CPU in, and the control and security maintains the same. You're seeing in configurations a server, more NIC cards, network interface cards. You're seeing more, storage cards.
All of those modularization in the server is more opportunities for Lattice. Each one of those boards is another opportunity for Lattice, hence you see our attach rate going up generation after generation. In fact, if you go to our last Investor Day, we actually gave some color on what, what that means from a, a dollar perspective, and what we said is that if you look at the prior generation, which is now ramping down, compare that to the generation that's ramping up. Take Sapphire Rapids or Genoa as an example. Not only is the attach rate increasing, but also the- because of the complexity of these servers, the a- average ASP of the FPGAs are selecting is slightly going up as well.
You combine both of those, you see about a 50% increase in dollar content from prior generation to the generation that's ramping up right now, which says that even if the number of servers in the market is flat, we still have the Lattice-specific growth driver that drives growth within the server market. Now, to our visibility, we work really close with multiple of our customers. They can be hyperscalers, OEMs, or even some of the ODMs as well, and we're engaged in the next-generation servers that are being designed. In fact, we're also engaged in the generation after that as well, and we continue to drive higher attach rate and more complex FPGAs are being leveraged as well, which drive a higher ASP.
So we see that trend continuing at least over the next 1-2 generations from what's being deployed today. For us, it's been a really good growth market, and you didn't ask the question about AI servers, but I'm gonna comment on that.
I, I was gonna go there. Well done. Carry on.
So Matt's gonna ask me, "What about AI servers?"-
Yeah.
What's your attach rate there?
Yeah.
So if you look at general purpose servers compared to AI servers, the attach rate is about at, at a minimum, the same as a general purpose server. Depending on the configuration of the AI server, it's actually much higher as well. So on average, if you look at our content of FPGAs and Lattice content and AI server, it's higher than that of a general purpose. So think of the baseline being equal or greater as far as an attach rate on the AI servers.
Got it. Yeah, well done. Anyway, from the audience, this doesn't just need to be us talking, so if anybody has questions out in the audience, just, Erica, I think there's a microphone coming somewhere. We're gonna time him running across here with the microphone.
Over here. Second row.
Well, thank you, first, Sherry and Esam, for coming. I was just curious, just given the downturn that we're hopefully almost through, as you think about the uptake of Avant, anything to note, and maybe looking back in history when you've launched products, maybe during similar periods in the cycle, is there any change in the way you think some of your customers are going to ramp, maybe with a little bit more caution in terms of how much they buy or how much they ramp, given maybe some uncertainties in their end markets?
Yeah, for Avant specifically, Avant for us is really exciting for a few reasons. Number one, we've demonstrated good growth of small FPGAs. Avant is additional market that we've opened up. We've increased our SAM by about $5.5 billion just because of Avant. And the Avant traction for us has been really positive. In fact, we had given the team internal goals of what to go after as far as design wins, and they've exceeded those goals, and we're really pleased with that initial revenue we got within Q4 of launch, which is a really good milestone as well.
From a customer traction perspective on Avant specifically, which is your question, it has not slowed down because of the cyclical nature of what the industry is going through. Customers and our customers still want to get their new products out to market. They want to differentiate in their products. They're looking for FPGAs that are lower power, smaller form factor, and give them the performance that they need. They want to get those products out as soon as possible. The current correction in the industry hasn't really slowed down that momentum with Avant, and with regards to the initial ramps as well, they're trying to get those products out to market. They're trying to get their customers to urge them. We're not seeing, at least today, any impact of this cycle on the momentum with Avant.
I think the other thing, just to add, is that I mean, the Avant, our customers asked for Avant.
Yeah.
which is why we developed it. So obviously they are interested in and really want the capability, and the other thing is that there's a 90% overlap with our existing customers for Avant, so that is also, you know, certainly helps the traction there.
One of the things I wanted to talk about is there's been some change, a little bit, at least, what people are projecting is going to be change in the FPGA space, right? There's some small, small companies that are sort of upstarts, but if you go back 30 years, there were four FPGA franchises, and there are still four FPGA franchises. Three of them have been acquired, but one of those is going to be spun back out into an independent company with Altera coming back outside of Intel. And there's been a lot of, I guess, conjecture as to what their strategy might be, and how Xilinx with AMD might react to that, et cetera.
So, just as you watch the market, despite of all, all the things that we hear, have you boots on the ground, have you seen anything really change? And what would need to change for the competition to really alter the momentum that you guys have?
Yeah, a few comments on that. Number 1 is you made a good point that it's still the same four franchises that have been out there for 40+ years. So as far as competing with each other, we've been competing with each other for 40 years, so that doesn't change. The FPGA market today has grown significantly since when I started in it, which was 35 years ago. I'm just aging myself right now. But if you look at the FPGA market, you can segment into 3 simple buckets. There's small FPGAs, mid-size FPGAs, and large FPGAs. There's just 3 areas, and they have slightly different characteristics to them. In the small FPGAs, we're talking about the largest customer segment. The most customers are serviced by small FPGAs.
As you go to the other extreme, in large FPGAs, it's a much fewer customers, but the ASPs of the large FPGAs, we're talking about $100s-$1,000s per unit. You contrast that with the small FPGA, you're talking about $1s-$10s, and in the mid, it's $10s-$100s. So there's a big difference between small, mid, and large. If you look at the competition that today specifically the two largest ones that focus on large FPGAs, that's their focus. They're building and innovating in large FPGAs. Lattice focus is in small and mid-range. That's where we see the need in the market, and that's where we have the most traction.
Whether one of them spins out, becomes independent, whether they be part of a data center or a computer or a processor company, from a competitive landscape, we don't see that really changing. We scenario plan, we take every competitor seriously, and I think the data shows that we've been gaining share and outgrowing the competition, and we'll continue to bring innovative products and differentiated products out there with our software solutions and continue to compete with the same companies we've been competing with for the last 41 years.
Totally makes sense. You hinted at it one of your answers earlier, to get to software. And there's two, I guess, parts of my question. I mean, my view, I guess, is that over the last 15 or 20 years, the two large FPGA players have been spending lots and lots of energy trying to make their FPGAs easier to program and easier to hook into classical computing systems, and-
Yeah
... they've made them more malleable for people that want to do work in those areas, which is an amazing contrast of what you would want to do to have low-power software. Like, those are two completely different things. So, I mean, you guys coming from a low-power environment, going into the mid-tier, maybe you could talk about how that software transition is going. Are the software stacks compatible? Are they with the same, you say, 90% customer overlap, how much software are they taking? How much software is overlaps between those customers and just what that progression looks like? Because it seems, to me, really hard to go from large FPGA software to low power. Talk about going from low-power FPGAs into the mid-tier on the software side.
Yeah. Software for us is a critical part of our overall strategy, and we've been investing more and more in our our software, specifically our software solution stacks, making it easier for customers to get to market faster and makes it stickier, as we talked about before. We did some analytics on our software solution attach rate, and we found that over 50% of the new design wins that we're going after have software attached to them. That says that customers are adopting it, they're appreciating it, they're liking it. And what we've also done is because we've got a more of a a pricing methodology based on value, we decided to take those analytics a bit further and take a look at the ASPs of devices that have the software attached versus devices that do not have the software attached.
We find that the ASP is significantly higher on those that are leveraging the software stacks, and that makes sense because we're selling value to our customers from, from an ASP perspective. But we continue to invest in these software solution stacks. We've got six that we've introduced into the market to date. We update those on an annual basis, sometimes on a biannual basis, and we continue to work with our customers on what are the capabilities. And think of these software solution stacks as pre-built applications, the most popular applications that our customers are implementing on FPGAs, and we do most of the work for them. We make it very easier for them, and we also simplify it for those that are FPGA experts and for those that are not FPGA experts, how to get the most out of our FPGA.
A couple minutes left. I think, Sherry, I wanted to talk about margins. You're gonna be going on a- I mean, the P&L is not the same P&L that you took over. Way not. Very different. So, well done. Thank you Sherri . But from here, as the company changes and morphs from a small FPGA company to across the mid-tier, how do we... I mean, you're gonna be- you were on at the SOI for a lot of products, now you're gonna be on more standard CMOS. There's a whole bunch more system components into doing mid-tier FPGAs than in the small tier space. Just walk us through what that means to margins. Should we just kind of, like, keep you at the current levels that you're at? What are the variables there?
Sure, sure. So, you know, when six years ago, when Esam and I started, and the company embarked upon our gross margin expansion strategy, you know, margins were, you know, in the mid- to high fifties.
Fifties.
And so we've increased it by about 1,300 basis points since that point. So, we're really pleased with that progress, but we're not done. You know, we continue to focus on our gross margin expansion strategy and, you know, a couple of elements there, that are part of that. One is certainly pricing our products for the value they provide. You know, Esam talked about the, you know, the number of new products that we've introduced. You know, we're, we're innovating and developing new products that our customers really want, and so they're willing to pay for that value, because our products help differentiate, their offerings, and so that's something that we've been able to do not only on our, pre-Nexus products, but we continue to do that on our Nexus products, and certainly with Avant.
Avant, the ASPs in Avant are 10-20 times those of Nexus. And so that pricing our products to the value they provide is very important, and we'll continue to focus on that with new products as well. Mix is another aspect there to gross margin. You know, six years ago, the consumer market segment was about a third of our revenue, today it's less than 10%. The consumer market segment was a very high volatile, low margin revenue segment. And so, you know, we intentionally shifted away from that, and that's contributed to our margin.
And then, you know, and with the core market segments that we have, industrial and automotive, and comms and compute, you know, 90% of our business roughly, you know, those are the market segments that you'd want to be in, right, in terms of driving innovation and really getting value for our products. And so those market segments are, you know, industrial automotive tends to be higher than our corporate average, and comms and compute is kind of around our corporate average. But mix has contributed. Now, on a quarterly basis, you can expect to see fluctuations because that can happen, right? Whether it's within the market segments or certain customers, applications, volumes, all that kind of thing, so that can certainly come into play.
Just to round out the gross margin expansion strategy, cost was the other element that we looked at, and we continue to focus on that because we wanna make sure, you know, that we get the best costs possible for our products with our suppliers, and so we continue to hold them accountable to yield efficiencies and things like that. So that's the other element there to our gross margin expansion strategy. Now, if you look back to, you know, where we are at our guide for Q2, 69% at the midpoint, you know, that we had indicated was down a little bit from sequentially from Q4 and into Q1 due to industrial and automotive softness.
That, as I mentioned earlier, that market segment has a higher gross margin typically than the corporate average. So, you know, there's the mixed interplay that can happen for gross margin. But having said that, you know, we expect it to continue to execute on our long-term model that we put out last year, which is the low 70s, and so that's something we continue to focus on. You know, with our gross margin expansion strategy, it's part of our DNA, continue pricing products for the value they provide, with a great number of product offerings that we have, you know, significantly increasing the cadence of those offerings that we've been doing at the company. You know, we continue to see opportunity to price our products for the value they provide.
We'll continue focusing on, you know, executing toward our long-term model from a gross margin perspective.
Wow! Very clear. Our handy dandy shot clock is expired on us. But thank you guys, and pass my thanks to Jim as well for the partnership over the years. It's been a remarkable turnaround of the company and exciting times ahead. Thank you.
Thank you, Matt.