Capital Markets Day Transcript - SMTC 2026-10-15

Operator

Welcome to Semtex Data Center Analyst Day 2026. Please welcome to the stage Semtex Senior Vice President, Investor Relations, Mitch Haas.

Mitchell J. Haws, Head of Investor Relations

Thank you. Good morning, everyone, and welcome to Semtex Data Center Analyst Day. We have a great program in store for you today. A lot of discussion today around where data center fits within our portfolio and how we're very well positioned for the future. We'll also talk about how that translates into our long-term financial goals. Looking at the agenda, we'll start with, sorry, Hong will kick us off with a discussion around the transformation of Semtech, where we've been on that journey and where we go from here. Following Hong's remarks, we have the pleasure of a very esteemed panel to talk about the future of data center networking, so we expect a lively discussion. Following that, Imran Shirazi, our general manager of Signal Integrity, will go into detail on the data center roadmap. Importantly, he'll talk about the legacy of our data center business and how that positions us for a very bright future. During lunch, if you see at the back, in addition to lunch, you'll have a chance to look at all the demos. Most of what we talk about today in terms of products will be featured in those demos, so a good opportunity to see the technology in action. Mark will begin at 1 o'clock with a discussion around the financial results for Q3. He'll talk about our long-term financial model and how that looks through the end of fiscal year 30. Following that, we'll open the floor to questions. The senior leadership team for Semtech will all be here, so a good chance to dig deep. A couple of housekeeping items, we will make forward-looking statements today. Those statements carry certain risks and uncertainties. The risks associated with that are available today on the website and as well as our SEC filings. In addition, during the course of the discussion, we will talk about non-GAAP financial measures. The reconciliations for those are on our website at investor.semtech.com. With that, thank you for joining us. Let's get started.

Mitchell J. Haws, Head of Investor Relations

The rise of AI has fueled massive growth in demand for data centers. with overall data center capex expected to exceed one trillion dollars in the years ahead driving a massive increase in power consumption power is the industry's next major constraint with a growing spent moving data rather than computing it making power efficient analog connectivity more critical than ever semtech is built to lead this moment we bring decades of experience of moving more data with less energy across copper and fiber with industry-leading analog and signal integrity expertise a proven supply chain advantage and the ability to scale that's just the start we're partnering closely with hyperscalers and ecosystem leaders globally to design what comes next shaping next generation architectures developing purpose-built solutions. As AI architectures accelerate, the content opportunity for our portfolio dramatically expands. Our data center business is driving substantial growth, strengthening our business mix, and opening the door to higher margins. Backed by increased investment in R&D, Semtech sits at the heart of the infrastructure powering the AI era. Because the digital future depends on better analog connectivity.

Operator

Please welcome to the stage Semtech President and CEO, Dr. Hong Ho.

Hong Hou, CEO

Good morning, everyone. Welcome to the Data Center Analyst Day. For those of you who come over from the OCP, welcome. It's very fitting for us to have the Data Center Analyst Day alongside with a great event of the OCP Global Summit. That's where the innovation was showcased. That's the direction we are heading to. I've been the CEO for Semtech slightly over two years, and during this time, you often ask us to hold an analyst day, our teaching day, to talk about where we differentiate and what are the acronyms of different innovations and technology. Here we go. The reason we waited for two years is because we were busy in transformation. But right now, I will say, and hopefully you see the proof, that Semtech is a different company. When I visit different sites, our employees often tell me, I have been working in this company for 15, 20 years. but I feel it's different, very, a lot of energy over the last two years. Our customers are telling us, how did we get to this point? Today, I wanted to walk you through our journey of SunTech rising, SunTech transforming, and SunTech exiling. So let's get started. But its inflection point is happening now, and we are fundamentally changing the company profile. With a sharper portfolio and after the divestiture announcement of Acera wireless module business that allow us to double down to have sharpened the focus on the core portfolio. you. And we have stronger growth momentum. The AI data center spending is increasing and the networking connectivity is bottleneck. And that is the area we are focusing on. That is the area we are doing well. And with the increased demand, we got tremendous booking momentum, we got record-level backlog that's given us visibility for many years in the future, so it's the right time for us to provide a multi-year financial guidance. And the early proof shows that we are on track to build a predictable, high-growth and high-margin business. I know I met many of you, if not all of you, over the last couple of years, but I still like to spend a few words on myself. I'm still as excited as leading this company as my day one. I feel very privileged to lead the next phase of some tech transformation and growth. I'm technical, had a technical degree in electrical engineering, and probably one of the few CEOs has worked in three national labs in three countries, China, Japan, and the U.S. And I also conducted research on photonics and communication at AT&T Baylor Laboratory in my early days. My industrial experience spans over communication, semiconductor, and data center. I have been moving around quite a bit, but that background really prepared me, I feel like, for this job at SumTech. The three parts of my experience are most relevant to what I do at SumTech. MCOR, at the time I was the CEO, started as director of technology, and we let the high power laser development and commercialization with supporting the all high power laser support probably 80 percent of the broadband infrastructure for cable tv you probably don't know that but that's why when we had the acquisition we know exactly where we're buying the second part when i was working at the intel to lead the cloud networking group that's where I get all the connections and get the familiarity on the data center requirement. Before I step in as the CEO, I was the independent director for a year. That prepared me to get to know the challenges and opportunities. All of this experience translates into the time. I was able to hit the ground and running without any learning curve, without a trial and error. so I feel fortunate to come to a right place at the right time bring the right skill set now a little bit more about some tech we got a long history we were founded in 1960 we became a public company in 1967 just to calibrate yourself that's one year prior to Intel was funding so when I engage with investors very often I heard oh I know some tech my ask is which part of the history do you know about some tech we started defense routed pivoted to analog and mixed signal and the acquisition in 2012 of Genom and Seclio really bringing in the market-leading asset. So that's our crown jewel. We did a little bit detour in 2023 of the acquisition of Sierra Wireless. At the wrong time, which stretched our balance sheet and diluted our focus. I'm not gonna dwell on that, but I don't want you to pretend that never happened. So I joined after that, And right at the time, the industry is investing in AI infrastructure tremendously. We analyze our portfolio. We feel we have the right to win. And we are embarking a journey of transformation for Semtech. Very established company, but it can have the renewed vision and focus. So today, we aspire to be the leading connectivity company of the AI era, focused on moving more data with less energy from the heart of the data center to the edge of the physical world. We calibrate our strategy, our action against this vision statement. So we embarked on a journey called Semtech Rising in three arcs and one story to transform the company. Semtech Rising to fix the fundamentals. We started that second week as I was a CEO, and we can declare now a mission accomplished for that phase. We are squarely in the Semtech Transforming. That is about to build more growth engines to accelerate our business growth. And our destiny, our goal is to be the leader of the industry in what we do, that's Semtech exiling. Again, there's three phases to support that vision. Let me talk about the Semtech rising. As I mentioned about, we embarked on this journey second week after I became a CEO. We didn't have time or luxury to do a long history study, but we know what's stressed us. We know what needs to be done. Our positioning was now clear. We were a mixed signal, analog, diverse company. That's not enough to get us focused on achieving our mission, achieving the alignment with the market opportunities. Balance sheet was stretched To say the least, we had a leverage ratio of 10.5x at a time, and that wouldn't allow us to do anything more aggressive. And then our employees' morale was low, and the engagement with the customers was lacking. So what we did on the strategic positioning is the delineated core and non-core and put our limited resources in a core area in R&D. And even for the core areas, we did a thorough review that we only prioritize a project that matters to the customers the most. And in the balance sheet strengthening, we did a number of transactions, and we brought the leverage ratio to slightly over one at this point, and we also prioritize our capital priorities in R&D, and being really nimble in G&A, and we were able to go on offense as we improve our balance sheet and financial capabilities. On the customer engagement, I'm going to have a slide to talk about that in more detail, but on the employee side, we communicate often and communicate at all levels through all-hands meetings, through monthly newsletters, through site visits, we reestablish trust and transparency, and I am very proud. Today we're going to be giving out of many numbers. This number I'm most proud of. All the engagement score from two years ago, I was handed over with a scorecard of 57% engagement to today, the 78%. That is really the thing that matters the most. If we did not have the employees aligned and to share the clarity and the vision, we wouldn't be here today so with that we were able to turn the highly leveraged company into a focused and aggressive data center pursuit so that to double click on one of the fundamental culture change it's a customer engagement process we provide ICs or key components now including photonics to the module manufacturers and then they in turn to supply the integrated product to hyperscalers and xpu companies in the past we basically engage with the module manufacturers on this chart and we do not really engage much with the definer the definational customer who design the data center who defined the architecture now it's different we engage at all levels by leveraging our signal the analog and mixed signal expertise and decades of right track record and experience we are viewed increasingly more by the hyperscalers as a technology provider as a technology partner. We can help to understand their pain points and by applying our differentiated technical analysis from simulation to purpose-built silicon to provide differentiated solutions. One way provide that what is doable and they can define their architecture and our solution by doing this engagement, it's almost automatically designed in. Of course, we engage with our direct customers very closely, and this helps us to accelerate the design in process, help us to translate into accelerated growth. As I said, we got early proof today. We are invited to the room in the early architecture design phase by the hyperscalers, by the XPU companies. That's really a great place to be, and we are increasingly getting trusted to be the solution provider. Some tech transforming, that's where we're squirreling in this phase. The aim is to build more growth engines. Now we have the financial capability. We want to double down and intensify the R&D investment. Just to give you a ballpark number, this year we're projecting to increase our R&D spending by 30% sequentially. In a core area, we're probably even more heavily indexed on the R&D. Why is that? To build more growth engines. Most of the projects, I would say, stemmed from the alignment with the customers in trying to solve their most pressing issues, not by the scientific curiosity. So we believe the R&D investment will inevitably develop a new product and generate new business opportunities. That feedback allows us to be able to invest more. So we are really in this good versus cycle to accelerate the business growth. We're in a really good place. And on the portfolio optimization, we did the first divestiture of non-core asset that allowed us to deliver a predictable margin profile. We're constantly looking at our portfolio, but we wanted to build more growth engines. The data center is today the largest one, and we have the lower, and we believe we can have more growth engine based on the current portfolio. So portfolio optimization is an ongoing journey. As we expand our portfolio and expand our ambition, we will identify more gaps in capabilities. But M&A focus is not to increase the scale of buying revenue, but more enhancing capability and apply the leverage. So we have some ideas and we constantly review that. that's a journey. Strengthening the winning culture. Our employees get an early taste of winning. That is great. And winning is indicative, winning is infectious, and winning is gratifying. Winning is our largest driver behind our employees. So you're going to have a chance to meet our application engineers during the break and roam around to see our demos, but we are aligned in our strategic imperative, growth-oriented, customer intimacy, innovation with differentiating solutions, do the best in whatever area we are in, and employee engagement. So we'll continue to strengthen that culture. And we are well underway for transformation of the company. To show a proof point that we are positioned at inflection, and Mark is going to be talking a lot more detail on our financial projections, but here I just wanted to show the historic three bars of the trend chart. It took us a year to grow the top line about mid-tenths from Q325 to Q326, but we're going to be having over 50% of growth from Q326 to Q327. That's from the updated guidance we're going to be giving. So our growth is accelerating Margin profile after divestiture Of the Sierra wireless module asset As expected to close in Q4 Our growth margin baseline is going to be 64% With tremendous tailwind of the mixed shift Into the high margin business And that is the floor, not the ceiling our earning is accelerating even more so because of the operation leverage when you have a larger scale without growing gna and that was translated into more earnings per share so we see the early proof um that inflection is happening it is not just a projection this is the data. This is a fact. Here, I said, you know, we are today, it's not a full-blown analyst day in reviewing our product and portfolios, but more data center centric is fitting to co-host with the OCP event, but also showing that in FY24, our data center revenue only accounts for about 7% of total revenue. This quarter, we expect the data center revenue account for about 40% of total revenue. It's not because the other part of the portfolio is shrinking. It's just because the data center growth is accelerating. We'll show you that the FY30 forecast later, Mark is going to present that and in that we're going to be having a data center revenue to be about two-thirds of total revenue with a much bigger number of the total consolidated revenue so we are still in the early innings of this growth even though this marked the inflection points and the portfolio optimization continues that's helped to bring the margin profile and also get a better bottom line. And our strategy is to become a leader in the data center. We may not be broad, but we want to be the best in everything we do, in every product we offer. It can be copper, it can be fiber, but the theme is to provide high bandwidth, low power solutions. We all see this chart, this is from recently, and also this trend chart is different. Maybe after this week, it's going to be revising a little bit, maybe up, maybe down, but you can't argue. There are no any of this level investment. It's unprecedented, and we're sitting in the front and center of this. In we engage with the customers, we were often told, you, the networking connectivity, it's a bottleneck. We have the opportunity to unleash almost as much as 2x of the compute power. What the customers want is a high bandwidth, low power, low latency, and high bandwidth density solutions. That's exactly what we do. So that's what we play in data center. We don't do cooling or power delivery, we don't do compute, or we don't do memory either, but we're just laser-focused on providing the best network connectivity solutions. So, talking about connectivity, you've heard the terminology many times, the different connectivity, but they all share the same requirement. Again, high bandwidth, lower power consumption, and a high bandwidth density and low latency, starting from scale across. This is the so-called, in the past, telecom, metro, but now it's proliferating into more on the same campus between different buildings. The reach is two kilometers and above. Through HIFO acquisition later on, I will discuss in more detail, we offer the industry leading gain chips to be integrated in tunable lasers and to be the high-end laser for Metro and scale across applications. Our laser is ready for concurrent light application as it's now and we are engaging with multiple customers on that. So we have some play in scale across that's only going to expand what is the most of the ramp we're experiencing. It's a scale out. So that is for interconnecting the Ethernet switch fabric in data center. And our solutions is incorporated in the fully retimed optical pluggable modules, LRO, LPO, AOC, you name it. And a lot of acronyms, and Emeron is going to be doing justice in his presentation to explain in more detail and lift up the hood and show what we have in there, what's our content in there. We're supporting 800 gig, we're ramping 1.6T, we're designing 3.2T. We provide multi-generational solutions, co-designing that with our customers. In front and center, that is all most of the activity at this point. Good thing is the volume compared to scale across. Estimated, you tell me, 15, 20, 25x of the volume. It's changing every day. Scale in, scale up first. That is even higher volume, 10x more. Scale up is to interconnect the servers, the switches, and at a rack level and cross racks within a cluster. The scale up, we provide a solution both copper and fiber. Copper through a redriver or linear equalizer, when it's integrated in a cable, The ACC is a great backplane interconnect solution, and that business is ramping with key hyperscalers. There are a lot of design activities of linear equalizer integrated on the board, because later on when MRON presented, you will see that signal integrity is going to be compromised when you transmit high data rate across the PCBs. The most elegant way to solve that problem is integrate redrivers, some use retimers, a little bit penalty on PowerSight, but you'll be able to clean up the signal and allow the signal to reach the other end without a distortion. Use copper when you can, use fiber when you need to. That has been our guiding principle and validated by Jensen in the GTC meeting. Otherwise, many of the analysts were asking me for a terminal value of copper. Don't put a terminal value yet. Copper is going to be playing in 224 gig racks. Copper is going to be even more playing in the 448 gig racks. I mentioned about the 3.2T design activities. That's for pluggable. The linear equalizer onboard, it's a tremendous activities in there. They see this as increasingly a very viable solution. Then the MPO, then CPO, then optical interposers, I assume those are really the key demos. I got a chance to walk on the floor yesterday for a couple hours at OCP. I probably already counted 15, 20 different demos of the MPOs. Our solution is there. If not they're today, we're tweaking the form factor, and it's going to be incorporated tomorrow. The reason for that is current MPO design is 200 gig per lane. That is an IP, that is performance is already validated through the pluggable transceivers. MPO, in a way, is LPO on steroid. You know, it's a highly packed form factor of LPO. So LPO performance is already validated, and we are in the front and center in providing solutions in there. So that is the market moving toward us. The market is moving toward analog, and we do nothing but analog solutions at this point. Scale in, you know, a lot of discussions, and people are really more forward-looking. when you run into the connectivity wall and the power wall and you can interconnect between different chips with the chiplets or optical interposer. So we are closely in tune with the customers on the different solutions. So that is a connectivity topology. And as the data rate increases just by providing the best PMD of physical media devices, TIAs and driver, it's not sufficient because people will give you the spec with a guard band. We don't have a lot of performance margin to give out. So that's why we need to expand our portfolio to be able to have optical components alongside with our electronic components. This co-optimization will allow us to provide a best chiplet, best chipset solution to the customers with reference design. Fitting in the theme of building additional engine, we did a couple acquisitions of Photonic asset. In March, we announced our acquisition of HIFO. It's a company going through different recognitions, but rooted with Caltech, Ortel, Lucent in 2000. Spent $3 billion by the company. When I was there, we bought this company from Lucent. And subsequently, we built high power lasers. and we leveraged 46 years of the heritage and we know that we are leaders in the optical components in providing the best power conversion efficiency. The design allows us to provide the best beam profile allows the low-loss coupling into the fiber, our waveguide, and also the noise figures. So this is really a great asset, a great extension. We are vertically integrated operating the fab in LA area from MOCVD EPI to fab processing to test and chip-on-carrier packaging. We can utilize the entire ecosystem to increase our capacity. Speaking of capacity, today we get more demand than we have the capacity. So we are in the process of increasing capacity. We project to increase capacity by 3, 4x by the end of this year and to serve the customers later on in different applications. One other way to increase the capacity is that we did another acquisition of a tonic asset in June of this year, D-Tech. So we brought on board the world-class designers for high-speed photodiode. diode and the acquisition also allow us to have the immediate access to 7,000 square feet clean room facility. So we got some pre-owned equipment, we're moving into this space. That's how we increase the capacity by 3-4X by the end of this year. There is a little different facility but within driving distance of 10-12 miles, and we are also focusing on design photo diodes and project to provide samples and demo in the upcoming OFC. So our photonics portfolio right now is still limited, but we are not planning to build everything for everyone. We wanted to anchor ourselves on the strengths and really proliferate strategically. For example, on laser side, I mentioned about our game chip is already class leading, provide the highest power in the tunable lasers. We have a solid customer base. With now HIFO inside of Semtech for six months, we are able to increase our customer base pretty significantly. As I mentioned already, our lasers in the same design already provide low noise, high power, and a lot narrow length width, which is our ideal for coherent light optical transceivers. We're engaging with multiple module manufacturers and then having our samples in their evaluation and qualification queue. Some of the early results indicating that we're better than any incumbent lasers they are using right now. The large market opportunity, the industry is still experiencing a lot of shortages is for IMDD, intensive modulated direct drive applications. You're hearing the laser shortages. Of course, we got a lot of interest in that areas. But we are, we don't have all the capacity to serve them. Instead, we use our differentiation. We use very high power lasers delivering power 200 to 250 milliwatt power CW lasers, which is different from, say, 30 milliwatt CW lasers. you have to use one laser per channel and eight lasers in optical transceivers. Now with our VHP CW laser, you can have one laser to drive eight channel in silicon photonics modulators. But we have to engage selectively because we wanted to engage with the customers. They really value the U.S. made and for security, for the supply chain security. And we use this as a platform to design for our future and develop OCI CW scale-up lasers working with partners. We provide UHP 300, 400 milliwatt lasers with a semiconductor optical amplifier. Together with their wavelength definition capability, we're able to provide DWDM lasers at a high power for CPO scale up. Again, we're not trying to be everything for everyone. In photonics, we are building strength outward and with the customer pulling us forward. The portfolio expansion can translate directly into the content increase in our business so here is a chart probably you can take a picture or you can remember that I mentioned about at 800 gig our content is single high single digit when the industry evolving into 3.2 T for example we'll be able to monetizing more with the same content component functionality with more complexity like TIA and drivers plus adding more content like CW lasers and photodiode and even modulators. This is really very important because most of the architecture shift in the future trend increases our content. That's why we're saying the market is moving toward us rather than away from us. So right now the content on 800 gig and through LPO increases, through MPO increase even more, and we are getting ready for the components to be inserted in the 3.2T generation. Okay, about the market size. So we use 650 group and And we'll have the founder and CEO of the 650 group, Alan Wakefield, who's going to be on the panel later on. So we use their data, and they are most in tune for the networking site. If you look at the slice in the teal color, that's the size of PMD, the TIA, and drivers. By the way, we adjust it only for our product, not the entire time. And, of course, this does not include any additional opportunities that we're evaluating from M&A point of view. But it does incorporate what's on our roadmap for organic expansion opportunities. Lasers and gain chips and SOAs, the addition of photonic components, you increase our SAM pretty significantly. I still feel that active copper has more opportunity than the market giving credit for because when you are going to a higher data rate, you just don't have other way to get to the other end without redriver or retimer. Either way, so in four years, you see the SAM can increase by 5x. That's tremendous. That is a great opportunity. For us, we grow revenue through three avenues. You know, we can simply ride with a wave, and the size and trend will benefit us. That's number one. Number two, the content increase. I mentioned about 10x potential from the previous slide. Number three, we aspire to continue our journey like what we have demonstrated on 800 gig TIA. gain market share on future generations so that give us a more growth driver as well so what give us a right to win why semtech i mentioned about a technical leadership and emiron is going to be talking more eloquently about that later on i'm not going to dwell on that we have the track record we have the data rather than proposal customer intimacy this engagement model works We will continue to gain customer trust, and we want it to be also easy to work with the customers. This is sentiment from the customer, unprompt. It's not what I wanted to do, and we really engage with the customers without arrogance, and we try to be problem-solving. Operations excellence. So we have started a co-planning process with our customers on a monthly basis. Now we proliferate that number to over 20. That gave us an insight not only for what immediately they need and also what they're planning in the future. So that allowed us to proactively design for capacity and on Tuesday we announced a multi-year strategic agreement with our great strategic partner, Tower Semiconductor. that agreement provides secure the capacity to support our business skills in the horizon we are going to be providing a forecast so we are also the hallmark of the operations for some tech team is to provide operations excellence all the time so those are the three pillars to support our vision I would be doing a disservice if I don't mention about spend a minute to talk about the LoRa because that's our second growth driver. And we also did some realignment with the market on the LoRa strategy. We focus on silicon, develop the best capability, lowest power, longest reach, and higher bandwidth, and provide the market enablement. So now we have the dual band to provide enhanced bandwidth. We have the lower plus provide enhanced the capability on another protocol. We have Amazon to address the mass market of consumer side through sidewalk. We have been traditionally been very strong in providing robust connectivity solutions for medical devices. The enablement through innovation forums through software development stack through these webinar and customer engagement to unleash the entire ecosystem so they work very hard alongside with us that allow us to proliferate the application verticals from traditional three to four to now 8, 9, and 10 or even more growing. That is a great growth driver. Early on I mentioned that we have planned to review the rest of our portfolio to continue to develop more depreciation, continue to develop better drivers. So what does that translate to? So Mark is going to be talking about this in a lot more detail. We're projecting by FY30 our revenue 3.2 billion dollars mostly driven by data center and the LoRa that translate into the consolidated growth rate of 38% compound just get that translate into the multiples 75% Kager translate into 9x growth in four years in revenue 30% Kager translating to nearly 3x of the growth of lower revenue from in four years and then consolidated is going to be about 3.5x. I wanted to provide a little bit more color on this projection and we have seen a lot of projection hockey stick but I wanted to stress that we heavily indexed on the near term because we have the visibility and confidence and conviction and so the FY 27 and FY 28 later on Mark is going to be providing for example data center site we project over 100% year-over-year growth the rest of this horizon we use the market taker but as I said we aspire to provide 50% higher growth rate than the market. So as for the bottom line, Mark is going to be discussing that in more detail. And now let me bring back our journey to our destination. Semtech rising, done. Semtech transforming within progress is accelerating. Where do we want you to bring the company to? Semtech exiling. We want you to be providing higher growth than the market, better profitability than the market to customers. We aspire to be their partner of the choice. To employees, we want them to be energized, motivated. And to investors, we thank you for the trust you place in us and in our team. team, and we want you to be handsomely rewarded. The last slide, the takeaway, Semtech is fundamentally different companies today. I'm now going to go through every bullet point, and hopefully you see the data, again, it's not the aspiration, and we are pivoting, and this is really, we are inflecting, and this is really in the early innings. Let me close on a personal note. When I started as a CEO a couple of years ago, I asked our employees and some of our investors in the audience, give me the benefit of doubt. I will not let you down. hopefully you see that the early proof we delivered some results but to me to my entire team to the board this is just the beginning the best of some tech is yet to come thank you very much so next on the agenda is a industry channel panel it's a ai connectivity impairments So it's my pleasure to introduce our moderator, Mark Edelstone, and the company's trusted advisor and also the chairman of Morgan Stanley Global Semiconductor Banking. Mark?

Mark Edelstone, Chairman

Good morning, everybody. Never underestimate the importance of leadership. That is one of my fond sayings. I've been doing this for a very long time, and to see Semtech today is quite impressive, quite amazing. Hong, in his earlier comments, said that knowing Semtech is a question of what period you're talking about. I've been doing this since 1989, and so I can vividly remember my discussions with Jack Poe, who was the CEO here from 1985 to 2003. and the difference in this company today under Hong's leadership is impressive, amazing and it's great to see. This company is in a great spot. It's an honor, Hong, to be able to do this for you today. We've got a great panel. I'll have them come in a minute but we're going to be able in the next hour to provide I think a pretty good perspective on what's going on in the market. Many of these things you'll know. Other things hope it'll be a deeper dive for you as well but But Hong and his team have actually put together some real subject matter experts around this topic, starting with Alan Weckl. Alan is the founder of 650 Group. He's basically been doing research for a couple of decades. He's been running this company since 2017. David Ramkoo is the head of infrastructure at Meta. So we'll get a pretty good perspective of what's going on with those folks that are spending lots of money on AI infra to deliver what's happening going forward in all the AI workloads and then Alexis Björlund who is deep technologist subject matter expert on this as well really great background currently now the chief strategy officer at General Catalyst which is a well-known VC investing in this sector so I'd like and then of course Hong is part of the panel as well so I'd like the three of them to please come up and join me on stage Okay. Thank you so much. So what I'd love to do, I gave just a very brief background of the three of you, but it'd be helpful, I think. You're such domain experts in things, and we're going to cover things from the market, technology, all those kinds of things. So it might be helpful to just quickly give a little bit more detail on your background. What are the things that you've done that make you impactful to basically help share inputs to this audience and then as you do that also the first question you could just do at the same time basically would be to just talk about the single biggest change you expect to see in data center networking here in the next five years so maybe starting with Alan and then we'll just go down the list and then we'll go from there yeah thanks so much for having me so I've been in research for over 20 years prior to that I did network design at Raytheon so My entire career has been in networking and recently in data center.

Speaker 6

And we track over 40 different markets from compute storage networking to semiconductor. So we're kind of, at the end of the day, unit counters when all is said and done out there. And the amount of change is just incredible, right? I think the way I think about this from a networking perspective is we're going through a great transformation in terms of speeds and feeds. But fundamentally, we're increasing the content as we go. So networking is kind of transforming twice, both from a higher speed, higher bandwidth perspective, and then also significantly more content. And that ultimately drives the networking market from tens of billions to hundreds of billions, and then eventually right to that kind of magical trillion-dollar market just for networking itself.

Mark Edelstone, Chairman

And biggest change?

Speaker 6

The biggest change is the pace of innovation, right? If we think about what's occurring in networking, we're going from one network to four networks in order to support AI, and that's something that historically we've never seen at an industry level before.

David Ramkoo, Other

Hi. I'm David Remku. I'm a part of Meta's infrastructure organization. My team is responsible for technical program management, product management, ecosystem and partnerships engagement. We also kind of lead the internal prioritization of your programs. We're part of the infrastructure team that's responsible for the physical hardware that we place in our data centers, and that ranges from servers and racks that we design that are based on merchant silicon, and every name you can think of is obviously represented there. We also spend a fair bit of effort on building our own silicon for multiple applications, including AI accelerators, network products, etc. Prior to joining Meta about four and a half years ago, I spent 18 years at Intel. My last role there was leading the infrastructure processing unit business within the Intel connectivity group.

Mark Edelstone, Chairman

And biggest change coming?

David Ramkoo, Other

Biggest change coming. How does the saying go? Predictions are hard, especially about the future. Probably the network evolves from being the plumbing of the data center to being the enabler of compute architectures that unlock new possibilities for our AI applications. Alexis?

Alexis Bjorlund, Other

Excellent. So thank you for having me here. So I currently serve as the chief strategy officer for General Catalyst, but have a deep career spanning NVIDIA, Meta, Broadcom, and Intel, and before that, some really excellent source photonics in the optics and networking space my entire career. So sorry about that. So I think one of the most important elements that we'll see coming down the pipe for disruptive of change, of course, is co-packaged optics, optics closer and closer to the silicon, whether that manifests itself in co-packaged optics, near-packaged optics. What we're seeing is an unlock of bandwidth that's required to enable next-generation architectures, as David's saying. So I think that's the big disruptive change that's coming up. But when we think about the evolving landscape of compute and memory architectures, it's all still bottlenecked and constrained by the network, and the network becomes the enabler to high-performance, heterogeneous compute systems of the future.

Mark Edelstone, Chairman

Fantastic. So, and everybody knows Hong already, and he's covered his materials. So, obviously, there's lots of amazing things happening in the sector. I think it's pretty clear that what's happening in networking in the data center is going to be very impactful in terms of how the rest of this whole AI deployment is going to play out. And I think we all know we're very, very early here. I mean, ChatGPT was only four years ago as an example. So tip the iceberg, in my opinion, despite what we're seeing in the marketplace today. And we'll talk maybe a little bit more about that, too. But just to set the stage a little bit. So, Alan, you just said trillion-dollar opportunity in networking over time, and we know that this year the hyperscalers are spending trillion dollars plus or minus. Just give us your sense of how you think it breaks out on the networking piece between scale-out, scale-across, scale-up, and then what do you see as the biggest, you know, the fastest growers within that?

Speaker 6

Yeah, so scale-out is the largest domain size today. That's the first market that's going to cross that $100 billion threshold. In the longer term, scale up will become the larger market and therefore is the fastest growing, but that does not take anything away from scale across, right? We're still in the early days on the locking scale across, whether that means building a building in a very large campus or connecting these campuses together. This is something that historically the hyperscalers didn't have to worry about in the volumes that we have but kind of agentic is unlocking that agents are by definition chatty and trying to engage with resources that sit in other locations so as we move from training to inference we really do get the unlock in scale across and that will have kind of its own growth rate but each of these is a very healthy market and you kind of add them up and that's where you get the significant TAM expansion fundamentally networking is becoming a larger part of CapEx as time goes forward.

Mark Edelstone, Chairman

Okay, fantastic. David, you're in a unique position here representing obviously what's happening in the players that are putting forth all the infrastructure and things not specific to meta because I recognize that puts you in a sensitive spot, but just kind of give your perspective as someone who is trying to put in all this infrastructure. What are the the sort of key things that you're focused on relative to seeing payback from the types of things that are required to basically make this all happen for the rest of the workloads that people are using?

David Ramkoo, Other

So if you're asking in the context of how we think about setting our budgets and projecting out our infrastructure, obviously we look at internal demand and we look at how the demand plays out over the years. Obviously, demand is plentiful, so the next key factor becomes power availability, and that generally becomes the gating factor of how much infrastructure you can add. Obviously, we look at TCO for the various solutions that we're implementing. We also try to map out the demand shape on training versus inference. As our models get bigger we need larger inference clusters on cluster size you're training you're generally trading training time you know versus you know the cluster size and the last but obviously it's become really important supply availability is really important you know so as we look out in time do we think that there is supply available for us to build to our ambitions you know so to kind of recap we think about power relative to our demand we think of our demand shape and And then we think of supply assurance and availability. There are some investments we make that happen well ahead of the actual NTI program that we deploy. So we know that we have to invest in capabilities before that actually intercepts a program, optics being an example of that. We've invested in optics and we've done optics research for a pretty long period in anticipation that when a product comes that intercepts, it gets integrated with that program. Otherwise, the duration of those deployments would get much longer.

Mark Edelstone, Chairman

Great. So maybe following on that thread a little bit, Alexis, you've got amazing experience. You've worked for the biggest and baddest companies effectively in AI. Your most recent was GDX, GJX4, NVIDIA, but before that, Meta, Broadcom. So you've seen a lot. But put yourselves in the shoes of someone who has to make these decisions around ROI. How do you see that today, recognizing that there's obviously a whole lot of front-loading that needs to take place, as David just talked about? How should investors think about the ROI on these investments today as we're thinking about the monetization of tokens and all the things that we're going to get out of AI?

Alexis Bjorlund, Other

Yeah, you know, obviously, I think the world is trying to figure out how to think about this. But the current way that we look at this is, you know, the return on investment of deploying AI infrastructure really is looking at the monetization models from currently we moved away from simple metrics of dollars per GPU per hour to dollars per token. And I think where the world will find itself is in dollars per outcome. So one of the most important things to focus on is AI diffusion into the enterprises, into sovereign applications, and understanding the outcomes that we can unlock with AI and those outcomes becoming value that we're able to attribute down to the investments. We're not there fully yet because we haven't been able to map out that full value chain. But what we are seeing is a tremendous rise in the use of AI, in the use of tokens, in the balance between the proprietary closed source models and the advent of open source models that are achieving very, you know, almost, you know, they're in rapid pace to achieve similar outcomes. So as we see the enterprises deploying a mixture of proprietary and open models, what it affords enterprises more choice in the cost structure of their deployments. So when we think about the deployment of more and more AI infrastructure, you know, I'm fairly bullish on that. So I think people are looking at this in terms, first of all, just in the AI infrastructure itself as a capital return. So you look at it in the land power shell, you look at how do we achieve an unlock of the bottlenecks that we have today. The bottleneck currently everybody's looking at is the access to powered land, to power itself so that we can scale with the ambition of the industry. And then when we look at an ROI at every different level of this, it's land power shell, typically people are looking for a 15% IRR. How do you bring that together? How do you execute that? When you're looking at deploying the AI infrastructure itself, it really moves up the stack. So I think we're going to see a tremendous value shift across the chain where value is accruing in certain parts of this today. It used to be accruing in the model and application and in the semiconductors. We're going to see that value shift across the chain as we continue you to deploy at massive numbers of gigawatt scale.

Mark Edelstone, Chairman

And maybe just as a follow-up, I know you're relatively new in your role at General Catalyst, but as the chief strategy officer, presumably you're thinking very hard about dollars that are going to be deployed by your funds. And so what's your observation as you look at it and look at the math behind it? Do the IRRs look like they are going to be achieved based on the opportunities that you see across the AI infra opportunity set?

Alexis Bjorlund, Other

Yeah, yes, you know, I think that there is a compelling case for the value to be achieved for, you know, as I said, at each layer of the stack, there's a different math, there's a different algorithm behind it. But one thing that is for sure is that the importance of compute and compute infrastructure as we see this evolution, whether the evolution, you know, As anything, this is a rising tide, and we'll see minor corrections back and forth. But the trend that's there, the ability for AI to deliver significant value and outcomes, we believe is going to be across the entire stack.

Mark Edelstone, Chairman

Great. For you, Hong, and for those in the audience, I sat in those seats for a very long period of time as a research analyst. I did that for 18 years I've now been a banker in the sector for 19 so kind of 37 years of seeing how things evolve and at Morgan Stanley we've looked at all these cycles you've probably seen research that we've done over the years of at every generation of compute starting with mainframes going to minis going to PCs client server you know wired internet mobile cloud all those things. It's typically lasted about a decade, and it's been an order of magnitude increase in terms of units, demand, all those kinds of things, returns. And now here in this AI era, it feels to me that it's well more than an order of magnitude coming. You laid out, Hong, a very robust view of what the world looks like in 2030. I agree with it. I bet you're probably low on those numbers is the Mark Gettlestone point of view. But we all recognize that things can happen beyond our control. And I know that as a leader here for this company, you, Mark, and us, your team are very focused on under-promising and over-delivering and making sure that you're looking for inflection points where things could change.

Hong Hou, CEO

So with that being said, what are the sort of the near-term demand signals that you're looking at to make sure that things are on track and making sure you can deliver when you look out just over a shorter period of time, like over the next few quarters as an example yeah no thank you mark yeah so we provide a multi-year guidance and as I said the near term we have much better visibility than booking momentum it's continued to accelerate over last four quarters we have been reaching the record level of backlog every quarter and I also triangulate and compare notes with our industry peers through say Morgan Stanley event and other industrial events what we see that is shared the same sentiment in the near term for the next two years the demand horizon is very solid and there is no single customer come in and hedging and said, well, can we dial down a little bit so I get more calls these days to can you help us to accelerate? They pull in some of the demand, even though we are scheduled for Q2 next year, and we wanted to have it earlier. And there was no sense we sensed that the industry, the channel is stacking with the inventories. so that gave us a conviction and a confidence for the near term the longer term certainly right now the market is a little choppy you know they look at all the different things but either way this is an unprecedented investment cycle and great opportunity and the AI user we know that today I would say the company wise we don't really know how to work without AI we're using AI in every area In our design, in our market analysis, in our node conversion, in our disposition of material yield analysis, and failure analysis, everything, in every area. So that gives us confidence that we are increasing the investment spending on the AI side. And I think we asked ourselves in the beginning, what is the return on investment? but now the baseline is with AI. We don't ask that question anymore. Simply put, we don't really know. It's not going to be going back to the days without AI.

Mark Edelstone, Chairman

Yeah, and it feels to me as well that just this opportunity here is so enormous as we just have to completely change the way that compute architectures have been done in the past. Jensen talks about AI factories and so on. And there's just gating factors here that prevent, in this cycle, things getting overheated versus what we've seen in prior cycles. So land shell power, wafers, there are just some natural gating things that prevent excesses getting built up, which, to me, if the demand holds, and I think we're still early on in the applications, one would think that this should go for a while. One of the things, though, Alan, that we hear from people is just, you know, the overall cash flows. You know, in some respects, people could argue the world is turned upside down. I love the way the world works today because all the companies that I think have just done more to transform our world in technology, being semi-companies now have pricing power. They've got variable cost business models, by and large. and a lot of the capex burden is now being burned you know being undertaken by their customers and they're obviously powerful like David's company and others but what's your view of you know just how long that can last and it sort of comes back a little bit to Alexis point of view on what the ROI is on those investments but what's your just high-level view of how long can we see this or are there issues that we should be focused on relative to hyperscalers neoclouds, enterprise, all the spend that's necessary to deliver the goodness that's coming down the road.

Speaker 6

I think it's very sustainable. Today we could say this is focused more on the hyperscaler space or hyperscaler plus neoclouds. What we're seeing on the early days of inference and agentic is an incredible unlock. Hong even mentioned that internally to himself. The big opportunity here is going to be the enterprise unlock and then shortly after that like the physical AI unlock and those are kind of tremendous cycles and if you look at the amount of investment relative to the unlock or relative to GDP or cash flow at a company level we're still very much in the early stages of that and I think that early stage plus sustainability plus what you mentioned right some natural gatekeeping items in terms of growth allow this to be a multi-year or say decade plus cycle to it fantastic maybe Maybe on that point then, Alexis, if we see any type of slowdown in just the cadence of models and creation and how that evolves, does that, you think, cause things to slow

Mark Edelstone, Chairman

down for networking or is inference so powerful that it's just going to carry the day?

Alexis Bjorlund, Other

I think we're seeing a massive conversion from data center and AI infrastructure being used from training to inference. And what we see working in the space today, or what I personally have visibility to, is an insatiable demand for compute, for inference. So I think even if the models were to slow down, we are just at the very beginning of this curve of people adopting AI and leveraging the current models, fine-tuning the models, fine-tuning them on their own custom data. we're starting to see real unlocks and the beauty of it is as the deployments become more efficient we're seeing even more demand so it is that you know Jevron's paradox of once people get access to the right model they distill it on their own data all of a sudden they're unlocking higher value and and being able to deploy further I mean we're I really believe we're just at the beginning of this so I don't think that that model slowdown would have an impact on the insatiable need for infrastructure.

Mark Edelstone, Chairman

Yeah, agreed. So, David, maybe on that point, because I would fully agree with what Alexa just said, how does that impact the way that you're basically building the network going forward from an inference demand perspective?

David Ramkoo, Other

When we think of the requirements of training versus inference, training requires a larger cluster that is more synchronous in its bandwidth requirements, inference is a bit more bursty, capacity can be a bit less predictable. You end up focusing more when it comes to inference more on scale up versus scale out. The other thing we think through is the larger context windows which people get accustomed to appreciating require greater memory bandwidth and the agentic Workloads obviously require a tighter coupling of GPU and CPU. So as we think about the capacity we put in place, we think about those inference requirements relative to training.

Mark Edelstone, Chairman

Great. Maybe for you, Hong, then, what's your sort of view, what do you see as being unique here about the networking block effectively compared with compute and memory? and how does it matter and where's the value creation that's going to come from here over time?

Hong Hou, CEO

Yeah, I'm sorry. Okay, I couldn't hear. Can you repeat a question?

Mark Edelstone, Chairman

It's really the question, what's unique about the networking block compared with compute and memory that's going to create the value around the networking opportunity that exists?

Hong Hou, CEO

So as I mentioned, And, you know, what we understand and often get asked by the customers that the compute power is plentiful in there and is innovating in an accelerated pace. Every 12 to 18 months, there's a new generation of processors coming out. But we do not, in a connectivity space, innovate as fast. So we become the bottleneck. And power is another key consideration. And every time the current connectivity infrastructure rely on DSPs, it almost feels like that's the crutches you have to use. If you can't design a clean signal path, you have DSP to do the cleanup job. But what we try to do is using an analog solution to provide a clear path for it. I do believe we can provide more value in the entire data center by innovating and provide a better solution in connectivity. I think we're getting a lot of, in a way, encouragement and pressure because of the accelerated innovation pace, because we are the choke point of the entire stack to monetize, to provide more value. So we can be enabling factor for the future.

Mark Edelstone, Chairman

Maybe when you think about that, Hong, as just a follow-up, it feels to me like there are so many different companies, the large ones as well as obviously lots of startups that are focused more on the compute side of things, and yet on this whole networking connectivity that's going to create a lot of value to make sure that we can break through the copper wall and through the memory wall, all those things, that the number of players that have the real core competencies to do this and what SumTech can do feels like a pretty small group of people. So maybe just talk a little bit about that. What you see is the real core competencies and the competitive dynamics around that vis-a-vis SumTech.

Hong Hou, CEO

Yeah, so that's a very good point. You look at the entire ecosystem, There seem to be more people work on the compute side and the networking side. I would say people start, the ecosystems just start realizing the criticalness of the connectivity. For example, Alexis once showed that Intel, back in 2016, launched the effort of the CPO. at a time we did a demo and the common comment is that oh we don't need that much power we don't need that much bandwidth so even though we had a technology but it didn't get commercialized until the compute really cranked out so much data for ai computing and the bandwidth density need to be increased dramatically now all of a sudden that become a trend I think the connectivity in a way is always you know at an Intel time we call adjacent it's not a front and center the front and center still the compute so that's probably part of the reason the entire eco startup ecosystem seem to be focusing on the compute more than the connectivity and the analog is kind of like a old-fashioned way and so now it's in a front stage because they recognize a power advantage for example the linear equalizer read driver provide the power saving by 90% compared to read timer and people did not believe it in the beginning. I said, wow, you know, really? Why? I mean, you know, now they start recognizing that. But you see more innovative companies as starting up, and I think the market is going to be driving the ecosystem to innovate. They're talking about the links by DLB lasers, by micro-vixel arrays, micro LEDs and even the RF. So people start paying more attention to the connectivity at this point.

Alexis Bjorlund, Other

I think it's a really great point, Hong, because you know we talk about data centers. I mean in the world that I'm living in where everything is measured in power. It's how many megawatts and then you back calculate from that how many accelerators, how many CPUs you could fit into that power envelope. And so we've seen this switch from simply model down model hardware co-design. It used to be hardware software co-design to model hardware co-design, model silicon, and now it's really, we're optimizing the entire system around power, and so some of these, you know, the analog approaches and anything that could save power or where we're innovative in that network, and then how do you build the orchestration layer of the software to be able to take into account the underlying layout of the compute system, the network, the network bandwidth, it's all being co-optimized. So it's a tremendous opportunity. When we see a constraint like power, it's, you know, really, truly grid-connected power where now everybody's building all the behind-the-meter power solutions just to unlock the capabilities. I mean, innovation is happening everywhere, and ideas that, you know, didn't meet the ROI for investment before are now being looked at again. So it's a renaissance across the network, I would say.

Mark Edelstone, Chairman

David, would you add anything to that? We've laid out where some of the constraints are. Obviously, a lot of innovation is going to happen to deal with those, but thinking about power, rack density, optics, anything else you'd add to what you just heard?

David Ramkoo, Other

If I think of the network power relative to everything else we use, it's probably a single-digit percentage number. Now, obviously we're deploying gigawatts so of total capacity so single digit ends up mattering but the real value of the the power of the benefit of optics in my view is from the architectural innovations that it enables so for example I'm somewhat if you guys had a chance to walk around the OCP floor you got to see that some of the racks now are getting a little bit bigger but that only allows you to extend the scale up domain so much if you want to get to a scale up domain of let's say 576 GPUs you need to find a way to connect eight racks of 72 GPUs together it's and for them to act as one optics is what what makes it possible so I think of optics as a lot of the value delivers is the added performance that those larger scale up domains as an example enabled that wouldn't otherwise be possible and or at the campus level at the at the cluster level we can now optics allows us to have these you know you know large clusters that are pretty significant distances or a lot more distant from each other than they used to be our you know so I think of the value of optics you know, significantly higher in that front. Our Prometheus cluster, our New Albany, Ohio, where we trained our latest model, Muse Wander, you've got buildings that are a couple of kilometers away. You've got buildings that are as much as 60 miles away that we're adding to the campus. It's a gigawatt cluster and optics allowed us to make that possible, make that, you know, act as one. Obviously, we had to put a lot of innovation in place to allow for that. We build a back-end aggregator site, which is basically a super spine. It allows us to aggregate regions of 20,000 or so GPUs into one. We have to build deep buffers so that traffic that's coming in from a couple of miles away doesn't get dropped. It gets merged and inserted with everything else. But again, in my view, the power is important, but the architectural capabilities and the performance you get from those is where a ton of the value comes from.

Mark Edelstone, Chairman

I'm glad to see there's just no complexity whatsoever in your business and the things you need to make decisions on. But maybe as you think about the things that are quite important to you or to other operators as well, how should people think about what to build internally and own internally as your real value prop versus what should come from the merchant market?

David Ramkoo, Other

Yeah, we basically think through about kind of three high-level questions if we decide to build versus buy. Do I have the ability to optimize the full stack, hardware, software, compiler, if I build something internally, and can I do that better? Is the volume sufficient to justify the NRE because these aren't small investments that we have to make? and then is there a TCO advantage as I as I look to scale out this internal solution generally we've taken the approach though that even when all those all three of those questions are positive we still want the fleet level to maintain a portfolio that includes both merchant silicon solutions as well as our own internal silicon solutions the reason being is we think that that fleet allows for a more resilient, you know, data center, and it allows us for, you know, for better options as we scale our solutions. So, but generally, those are kind of the three high-level questions on how we think on make versus buy.

Mark Edelstone, Chairman

Great. Maybe, Alan, coming back to you, because we just talked a little bit about power, and obviously that's a big issue, and people now are talking about compute as a function of power, and the numbers I keep hearing just keep changing seems like every quarter, every year, in terms of, like, what is the amount of spend per gigawatt of capacity, effectively? So from the research that you're doing, what do you think that is today, and does that change in the future? Because I keep, it's like every time I hear someone tell me, you know, sitting in, like, David's seat or someone else's seat, it's a pretty high number. So just sort of cost per gigawatt to deploy.

Speaker 6

Yeah. So I think that's really operator-specific, right? it's a very different answer if you're a GPU as a service versus more of like a search or social platform and what you're trying to do with those workloads but networking right now is kind of in that we'll call it 12 to 15 percent range and that's going to go towards the 20 percent range in terms of spend per gigawatt and ultimately networking is that tool that unlocks everything that we said like networking is the tool that unlocks power it allows you to go to that additional power source. Networking is a tool that allows you to unlock context windows, right? Pool that memory together. So networking is a required tool in all of these unlocks. It's like the only universal tool that lets us solve all these problems.

Mark Edelstone, Chairman

But just so we get the denominator right, what's your view of total dollars per gigawatt all in? Compute, you know, everything that's going in to basically make that happen.

Speaker 6

Yeah, so total dollars per gigawatt is somewhere around $35 to $50 billion depending on what you're doing, build versus buy, et cetera. And I would expect that to increase with time, not decrease with time.

Mark Edelstone, Chairman

Yeah. I just heard someone yesterday tell me $80 billion. So it's like the numbers just keep going up. And obviously, that's in part because complexity is going up as well. And You know, this stuff is not inexpensive, for sure, either. So, Hong, kind of your perspective of proprietary versus open networks, how should we think about that opportunity set of what things are going to be just open standards versus things that might be more proprietary as it relates to the networking piece?

Hong Hou, CEO

Yeah, so it is great that, you know, right now I think we are doing both the standards and the purpose build. So OCP, by the way, David, on his spare time, he's leading the OCP Open Compute Project as a chairman. So I really appreciate you break away from the show and come to join the panel. But the standards define some common ground so that the innovation ecosystem can build and innovate in the most efficient way. But with the accelerated pace of innovation that we talked about early on, you have to engage to allow some purpose build. And so the engagement with hyperscalers, they have the scale. They move very fast. They know exactly what they need for their architecture, for their networking. So that's what we found. There's a new way of engagement. It's a combination of both standards and early engagement. So that's a part, as I talked about, the engagement model we changed through the Semtech rising phase, is that having the end-customer engagement to define what the future generation, And for example, right now, the XPO and MPO, the MPO probably still doesn't have standards yet. The standard pinout and all of this connectivity requirement. But people started innovating already. You know, there may be some form factor will become the de facto standard through this course. But you cannot sit back and waiting for industries to argue out and define a common standard. and then you develop, and that will be just too late. So that's a challenge for us. That's also excitement for us to see that connectivity is finally not just a conduit to connect something. It is an enabling factor for the uptime and efficiency of the AI data center.

Mark Edelstone, Chairman

So maybe, Alexis, turning to you again, and you highlighted a little bit on this before, but maybe just give us a level set of where we are today in the journey between pluggables going to NPO, to CPO, XPO, and so on. Where are we in that path? As we get to CPO in particular, what are going to be the key things that you think will determine success for the suppliers? Because it's going to take a lot to get embedded altogether.

Alexis Bjorlund, Other

Yeah, I mean, I think, you know, this really comes down from my perspective, it's in on having operated large scale infrastructure, it really comes down to an operational metric, what people are trying to optimize for is the uptime and the reliability of the infrastructure itself. So as we look at the decision points, and we look at where the failures are in the data center itself you know there's a obviously the entire industry works on pluggable optics today but we're seeing the switch over to co-packaged optics um i know that there's a big debate on co-packaged optics versus npo but when you step back from that both of them require removing the entire blade to resolve so my personal view is the leapfrog to co-packaged optics because it It really comes down to mean time to repair, blast radius, mean time to repair, and how do you replace that blade. So I think that the ultimate reliability of a Co-packaged Optics solution will be higher, and the goal of the industry is to bring that to a standard silicon level reliability. And then you still are using these, the optical light source is still a pluggable solution. So I think that that light source remains field repairable. And then it just comes down to the architectural decision of how many light sources per pluggable unit. So not to make it too technical, but I really, when you step back, I do see the evolution from pluggable to co-packaged because it's a significant improvement and it's at least a 30% improvement in power consumption. And then the signal integrity and the performance is extremely high. Now there is, I think there will be people who choose to go to the NPO route because it's more comfortable. But, you know, we'll see every different hyperscaler adopt a different solution on that. But ultimately, you know, that would be my own personal view.

Hong Hou, CEO

Yeah.

Mark Edelstone, Chairman

Did you want to add something, Hong?

Hong Hou, CEO

Yeah, I agree. I mean, ultimately, the CPO provides the best power efficiency, the best integratability, and also the stability. People have the, you know, kind of, in a way, misconception. Oh, there's a serviceability of reliability. What if the lasers having reliability issues? Well, traditionally, laser has been the root of all evils. You know, if the optical transceiver failed, nine out of ten times it's because of lasers. This stage is different. Laser reliability is not an issue. What Alexis mentioned about the uptime, the reliability, it's more the network, the link flaps. And, you know, I hate to say this, when you have the DSP, when you do not test out the corner conditions in the network, it's prone to link flaps. So, CPO, you don't bring the optics so close to the XPU, you don't have, you don't need a DSP. you don't have the chance to kind of getting that complication integrated. So I think that's great. MPO, in the meantime, is an interim open ecosystem play. So when CPO, they are establishing the entire manufacturing ecosystem, because based on the silicon photonics, this is not as mature as a digital logic from design, from packaging, ecosystem. But the industry is building that, and they're catching up so quickly. I have no doubt that CPO is going to happen in the future. MPO, in the meantime, provides viable interim solutions, allow different hyperscalers to get their customized XPU interconnected while they don't have an industry standard in defining pin-out and mechanical keep-out zones.

Mark Edelstone, Chairman

Yeah, agreed. And it feels to me like next five years, CPO is likely to be one of the most disruptive things we see here in infrastructure. Maybe, Alan, from your point of view, and I recognize it's sort of changing, but where do you think the knee of the curve is for CPO deployment? Like, when does initial ramp really happen, and when does it just catch fire?

Speaker 6

Yeah, the initial ramp is now, right? There is CPO in production and in, I'd say, decent or significant volumes beyond proof of concept. the inflection is going to be second half of 28 or into 29 where that inflects there's kind of two bottlenecks there the first is manufacturing CPO is new and it takes a lot of new tools in order to get that through manufacturing test qualification that's new the industry needs to grow into that and then secondarily you've got a time into these road maps whether that's on the or call network ASIC or the XPU GPU, right? And so that kind of creates a question like, do you go NPO to CPO? Do you jump right to CPO? That's really going to be a personal decision of each hyperscaler or each accelerator. But the inflection is second half of 28. And as you mentioned, like, it's going to be highly disruptive, but in a massive opportunity way.

Mark Edelstone, Chairman

Yeah. So David, congrats to you on OCP, given your role. It seems like it's been just a home run and you've outgrown this area. so I understand it's moving in the future, but it seems like one of the key things coming out of the show has been micro-vixels versus other approaches. So do you have a view on that, micro-vixels, micro-LED? How do you think that ultimately plays out? I recognize that it's still early, but decisions are needing to be made now for these ramps.

David Ramkoo, Other

Decisions are being made now, and if I kind of put my OCP hat on, generally our approach is to allow the ecosystem to debate and arrive at a solution that there is a greater alignment on. It's probably not the way to go, it's probably not the fastest way to get to what ends up being a standard, but we find that it's the most enduring way because you've allowed the technical debate to take place and you've allowed the voices of support and opposition to bear out.

Mark Edelstone, Chairman

And OCP is obviously really well suited to allow for that to happen as various work streams of companies come together collaborate and ultimately you know turn that into actual products that enter the market so very very nice consensus building thought process Alexis is a deep technologist or Han maybe the two of you what's your when you think of it from just a pure technical perspective of solving all the challenges that would exist which do you think is gonna be which thing wins out in the end, recognizing we'll coalesce people's perspectives, but the technologists knew where would you think this shakes out? Microvixels and micro LEDs is an example.

Hong Hou, CEO

Microvixels and micro LEDs. The slow and wide and for scale up.

Alexis Bjorlund, Other

So now they seem to be the recent trend is the argument is micro LEDs you leverage the consumer electronics industry the micro LED for display you do the slow and wide you know I know your opinion on that yeah sure I mean so yeah so I yeah I mean as a pure technologist as an individual you know I believe in the micro pixel or pixel approach over a micro LED approach but that's just because of the signal integrity and the high bandwidth communications so also when you think about the challenges of the connector and how large the connectors need to be you know it's really about system design but that's you know personally I also lever towards the the higher bandwidth solutions higher bandwidth less connections because so many fibers, so much connection. I've just personally seen inside the data centers and in the racks how difficult that is. But, you know, what is your opinion, Hong?

Hong Hou, CEO

Yeah. No, Alexis, I totally agree. And it's slow and wide and need to be reasonably fast. If it's too slow, you're adding a reverse gearbox, you're getting the native signal converted back to slow signal, and the other side to do the other conversion, this inevitably adding more power consumption kind of in a way defeat the purpose and but of course the better to be fast and white you may need that and that is it was really the OCI it's promised to deliver this eight channel DWD later on 16 channel even had a promise to do 32 channel each one go with a 64 gigabit or 128 gigabit per second that will provide a high bandwidth and a high bandwidth density too but i do see that industry thinking about all the different ways to innovate not only for performance for availability for the cost for reliability those are really the thing that ocp at a consortium, you know, to try to promote, right? Not just focusing on technical aspect, but the whole totality of the supply as a viable solution.

Mark Edelstone, Chairman

Yeah, okay. Maybe a question for both Alan and David here. As we look at the transitions that are happening here from like 400 gig to 800 gig to 1.6T, 3.2T, the cadence is shrinking. And so that's great. We're ramping stuff faster than ever before. Is that sustainable, or does something ultimately have to give, maybe from the market perspective and then from the user's perspective, David?

Speaker 6

Yeah, I'd say from a market perspective, it is sustainable, right? 800 gig is over 2x the size of 400 gig in terms of volume, and we can expect that at 1.6 will be a 2x plus the volume size there. So the market is very much able to absorb those types of volumes, but we are running into a little bit of a dilemma wall there. We've more or less maxed out the radix from a switch perspective, and we've maxed out the beachhead from an ASIC perspective inside the server. So that requires the more intuitive ways. That's why there's so much excitement over microvixels this year, right? It solves some of those beachhead dilemmas. So the market can absorb it, and the volumes are absolutely there with the cadence.

David Ramkoo, Other

The cautionary tale is if you're at 1.6T, how much 400 gig you're going to have so you ultimately want to make sure that you have a home for those older technologies as we go through these faster adoption curves yep and from a user perspective yeah i mean so i'm obviously not going to answer uh the sustainable uh question uh but the way i think about it is that evolution of the network is kind of matching the evolution of compute and that evolution of compute is largely matching what our uh you know kind of needs are So if I think of it from that perspective, it is challenging that you're now seeing the cadence accelerate, and it's accelerating on all fronts, on all of our NPI programs. You're kind of bringing, you know, the next generation faster together, which means your quality cycles either have to take less time or they have to be parallelized. Most often they are parallelized. You're deploying one generation as you're working on the next generation as you're kind of planning the third generation. And so it is a bit, you know, challenging from that perspective. If you look at how hyperscaler CAPEX has evolved just in the last, you know, four or five years, you can assume that the number of programs, the complexity of those programs has increased commensurate with that, which, you know, is kind of a key indicator that, you know, the compute and the performance that compute delivers, you know, grows. and obviously it's really important for the speeds in the network and the bandwidth in the network to evolve. Like I said, I'm not going to make a forward-looking statement on that, but, you know, it kind of matches the needs of the data center infrastructure.

Mark Edelstone, Chairman

Yeah, great. So we just have a minute or so left, and it's been a great discussion. There's obviously lots of complexity, lots of opportunity here ahead in this whole opportunity set for networking to drive AI infrastructure and all the benefits that it's going to deliver in the future. Like I said, it's a complex world. There's lots of, you know, the market's concerned. What's the one just assumption you think that the market just has wrong today about, because to me it's obvious where things are headed, but the market doesn't always see it the same way. So what's the one thing you think is an assumption people are making that just doesn't hold water. Starting with Alan.

Speaker 6

Yeah, I think it's that these are binary answers, right? We're going to get to CPO. There's no doubt there, and it's going to happen, I'd say, roughly in that five-year horizon. But the misconception is that these things are binary. Copper goes away. Copper's not going away. If we do micro-vixels, we're not getting rid of transceivers. So these aren't going to be binary events. It's the right network for the right part of the server or the right part of the memory, and that ultimately leads to a lot of markets running in parallel.

Mark Edelstone, Chairman

David?

David Ramkoo, Other

I'm humble enough not to make a statement about what the market is wrong. I generally observe the market to learn from, and I over time find that it's actually more pressing than most forward-looking statements out there. If I were to add one area where I think it's probably underappreciated by, you know, the outsiders, is the amount of innovation that's happening in the hardware space right now at all levels. There is more emerging companies, there is more startups, you know, many of them have fantastic ideas they're working on, and the rate of innovation is kind of unmatched to anything I've seen in the past 25 years I've been in this space. And I think that's generally underappreciated by the outsiders, those outside looking in.

Alexis Bjorlund, Other

So I agree with what both of you have said. I mean, I would take it just at a slightly different angle is I'm an incredible optimist for the power of AI for human outcomes. You know, I think this will be an era of incredible abundance that we're about to unlock. so it may not be what the market's getting wrong, but I think that there's a lot of fear in the market that could really be converted to optimism. We're at the era and at the precipice of an incredible time, and it's going to require an incredible amount of compute and compute access, so just like, and it's the opportunity to, I fundamentally believe it's an opportunity for all of humankind globally to seize this moment and really unlock the power of AI. So that's more of a wish on my part, but I get to operate at the very frontier of AI. I see where it can go, and so I really am very bullish because of that, and I think that we're going to unlock this moment. This moment will be really better appreciated maybe in the coming year.

Hong Hou, CEO

All right, Hong, you get the last word. So I think the disconnect, what I see is today's assumption of sustainable or not, is based on the technologies today. But we continue to innovate. For example, you're going to be hearing this afternoon, XPO, increased density of 4X power efficiency can be improved. So the cost of building data centers, the power consumption, and the labor and the footprint you need, it's all going to be revolutionized. People have now factored that into their model and look at the macroeconomic situation. Another thing is a disconnect between the macro concerns and really we are in a trench building the product and we have the great visibility and the daily drill we have to go through in expediting. So I hope sometimes it will be converged somewhere. But I'm very excited about this. I do believe, as I said, the best is yet to come.

Mark Edelstone, Chairman

Yeah, I couldn't agree more. And you can't change the world in a straight line. But I think there's this tremendous opportunity ahead. I want to thank the panel for just great insights. Thank you all for being here as well.

Hong Hou, CEO

Thank you.

Operator

Please welcome to the stage Senior Vice President and General Manager of Signal Integrity Products, Imran Shirazi.

Imran Sherazi, Other

Good morning, ladies and gentlemen, and welcome to the Data Center Deep Dive. My name is Imran Shirazi, and I'm the General Manager of the Signal Integrity Products Group. And I'm really excited to be up here this morning to share with you our vision for Signal Integrity in the Data Center and why we believe passionately that the market is moving towards us. I've had the pleasure of meeting some of you before, but I see lots of new faces. So welcome, everybody. Thank you for attending. And I look forward to meeting with you afterwards, perhaps over lunch or maybe at the demo stations at the back of the room here. I'd like to begin today by sharing with you our incredible history in signal integrity. Many of our team worked together in the late 1990s in the telecom era, developing TIAs at 50 megabits for OC1, which was the first optical standard. The business in its current form came together in 2002, and the first success was in the telecom market at 2.5 gig and then at 10 gig. but we quickly pivoted to 10 gig Ethernet and I still remember when we hit ten million dollars in revenue we were a small group back then and the lead designers some of whom are still with us was so thrilled that their designs were competing in the open market winning sockets and being installed in networks that were becoming the internet we used to tell our friends and our families that their emails were sent and received over our chips. We went on from 10 gig ethernet to address fiber to the home, then 4G wireless, and then the cloud data center at 4 by 25 gig. We developed all these product generations, but it was the passion for winning, for seeing their work become part of the network that really drove those designers to the innovation and success that we've built the business on. As the line rates went up, we became the analog signal integrity pioneers. And I don't say that lightly. Signal integrity is a dark art. That's why it's called a dark art. You won't find it in textbooks. Groups like ours were developing the art through generation after generation of product, constantly pushing the boundaries and learning hard-won lessons as we grew new grads came in they were coached trained taught that dark art by the senior engineers and then they too became part of that cult of signal integrity and here we are now delivering 200 gig at scale introducing 300 gig and designing 400 gig. When you look at this graph, you'll soon realize that the AI data center is literally what our business was built for. And that's why we passionately believe that the market is moving towards us. We talk a lot about signal integrity. I'll talk a lot about it here. So I thought it would be a good idea to give you an introduction to some of the principles of signal integrity. What is signal integrity? It's about impairments. It's a set of impairments that impact a signal, especially as that signal propagates across a channel, and how we deal with those impairments. So let's have a look at one of those impairments, attenuation. How do we think of attenuation? I'm speaking up here and you can hear my voice quite clearly but if I start to move away from you my voice will get quieter and quieter and quieter that's attenuation from my from my position here to your ears is the channel and the signal is my voice and the further I am away from you the more my voice will be impaired and that's exactly what happens to a digital signal in an AI data center today it suffers attenuation the further it propagates across a channel and what you see in the diagram is a healthy digital signal to start with a transfer function where there's some attenuation and a very diminished signal at the end there. So low that it doesn't cross the decision threshold. And so it doesn't register as a signal at all, just a series of zeros. That's attenuation. How do we deal with it? With gain, like a microphone. We add gain to the signal, bring the signal amplitude back up so it can exceed the decision threshold, and those zeros and ones can be read correctly. Let's look at a second impairment, bandwidth limitation. You're all quite familiar with bandwidth limitation. Every time you're on hold on a call with a call center of some kind, you listen to music and it doesn't really sound like music because that channel, that telephone line channel, is very heavily bandwidth limited. The human ear can hear from 20 Hertz to 20 kilohertz, but that telephone channel is limited from 300 Hertz to 3,400 Hertz. Just a narrow slice of the audio spectrum, a narrow slice of that high fidelity music spectrum. So it doesn't even sound like music. It sounds flat. The lows aren't there. The highs are heavily attenuated. It doesn't really sound like music. It's almost as if the words are smearing into each other and that's what you see here with a digital transmission. The effect of bandwidth limitation on that digital bit that passes through the channel is for the rising edge and the falling edges to spread out because the high frequency content has been attenuated, bandwidth limitation, it's the high frequencies that don't make it through and so you can't get that nice square wave. Instead, you get this spread out pulse, and it's reduced in amplitude because the high frequency content is gone. And so some of that energy is lost. And as you can see, what started off as a 0001 isn't really being picked up. That force 1 isn't being recognized by the receiver due to that bandwidth limitation. And what do we do to overcome bandwidth limitation, we use linear equalization. Apply EQ to the signal in order to prop up that curve that you see there to get more gain at those higher frequencies in order to allow those high frequencies to pass through and maintain the alacrity of the signal. Let's look at a third impairment, non-linearity. This isn't a channel impairment, this is an impairment that occurs in active devices. Devices that don't have sufficient linearity. What's a good example of that? You take a guitar, an electric guitar, and an amplifier. As you turn up the volume on the electric guitar, the volume from the amplifier increases until you overload that input stage to the amp. And that's when you get distortion. And of course guitarists love to whack up that volume, get that lovely distortion. We all enjoy it, but not in signal integrity. Here we don't want that non-linearity. We want our devices to operate in the linear region. And how do we do that? How do we ensure that? We use automatic gain control. So we control that gain. We make sure we've got enough gain for small signals, but we reduce that gain when the signals are large. We can also use pre-compensation. Finally, let's look at noise. Noise occurs in every system and AI data centers are not an exception to that rule. You get noise from the power supplies, you get noise from adjacent racks, you get noise from the fans in the back of the power supplies, and you get noise from adjacent lanes. Every channel, every lane that's transmitting a high-speed signal is radiating RF noise. And the channel that we are sending our signal down is susceptible to that RF noise. And so what you see is noise on the ones level, noise on the zeros level. How do we think about noise? Again, I'll go back to my example of my voice. You can hear me quite clearly right now. But if somebody were to start operating a jackhammer by the side of the stage here. I might get drowned out. That noise level, if it's greater than my voice, if the noise is greater than the signal, that's a low signal to noise ratio. My voice is lower than the noise. Ideally, that noise wouldn't be there. You can hear the fans humming in the back there from that 64-port switch, if that were the only noise, my voice can overcome that noise, a high signal-to-noise ratio. And so every signal is susceptible to that noise. If that noise is very significant, you might see that noise on the ones level or that noise on the zeros level being so significant that it exceeds the decision threshold. And that would result in an erroneous bit. What's an erroneous bit? We talk about BERs. 1E minus 9, 1E minus 12. An erroneous bit is a BER. 1E minus 9 is 1 in 1 billion bits. One erroneous bit out of 1 billion bits transmitted. 1E minus 12, one erroneous bit out of 1 trillion bits transmitted. How do we deal with noise? We use low noise design for a start. Secondly, if the noise isn't in band, if it's not at the same frequency as the signal, we can filter that out. That's easy. It's the noise that is in band that's on that signal that is the real challenge when it comes to managing signal integrity. I hope that's been helpful for you. Quick overview. It's obviously a lot more complex than that but let's see how some of these principles apply in a real world data center application so here we're showing a GPU ASIC talking to a switch ASIC at the far end of the screen that I'm sure you can imagine today is a very common use case in a in a data center the GPU ASIC will output a nice clean waveform as you can see there. Maybe some noise gets added and so that signal integrity gauge that you can see below the signal starts off firmly in the green. It's a nice clean signal. Some noise gets added. That signal integrity gauge moves to the right but it's still in the green. Bandwidth limitations of the channel as we discussed. That signal integrity gauge is now moving into the orange zone. This is starting to get challenging for the switch ASIC at the far end to be able to resolve with a good bit error rate. Finally, attenuation. Now that signal integrity gauge has moved right into the red and if you compare that signal trace in the red box with the original transmitted signal, they barely resemble each other. And so the switch ASIC at the far end, there's more distance to go yet, is going to have a real challenge in resolving that signal with a good bit error rate. So we introduce a linear re-driver, a Semtech IC, one of our copper edge ICs. That linear re-driver works its signal integrity magic. Some of the approaches we discussed on the last slide, plus many more and you get a nice restored output signal. It's not perfect but it's restored and it can move, it can propagate some distance down the channel, be received by that switch ASIC, get resolved without bit errors. Let's look at another example. We've all heard about retimers. Here I'm showing at the top, a re-timer example, and at the bottom, a linear re-driver example. What's the difference? What does a re-timer do in that chain? You may have already spotted it, but I'll highlight it here. The re-timer has effectively reset all the impairments before it, all the impairments of the channel. It's recreated that signal. If you look at the technology inside a re-timer, inside a DSP. It's essentially the CERDES technology that's used in the ASIC where the signal originally originated from. So it can completely reset that signal, recreate it perfectly, and send it down the channel. How do we think about that? Retimers versus linear redrivers. Picture an example of a flight from Sydney to London. The re-timer example would involve a stopover in Singapore. The flight would land, that's latency, would have to take off again, consuming a lot of fuel, that's power, airport charges, that's cost, the footprint, versus the linear re-driver example is like a light, lighter aircraft taking off from Sydney, flying direct to London. How? It gets a boost along the way, a strong tailwind. And if you think about that boost, it gives you the opportunity to maybe take off a little lighter. In other words, an opportunity for the host ASIC to perhaps not do as much heavy lifting as might have been required in the past. And that's the beauty of linear signal integrity and why linear solutions are becoming more prevalent in the data center today. The way we think of it is like this. Linear when you can and retimed when you must. Thank you, Jensen. Let's look at signal integrity by function. understand some of our key products. Linear redrivers, we just talked about those, we saw how they work, and essentially they extend the link budget of an ASIC. That's the value proposition. At their core they do one job, to invert or undo the effects of the channel. We just talked about those various effects of the channel, that's what a linear redriver does, does it very efficiently, a very elegant solution. If we look at the function of a TIA, a TIA takes the signal from an optical fiber that's been converted by a photodiode into a tiny electrical current on the order of microamps. And that current has to be converted to a voltage and that's transimpedance, the conversion of that current to a voltage. That's what a transimpedance amplifier does. It undertakes that conversion while maintaining very low noise, very high bandwidth. Pardon me, not very high bandwidth actually. Some of my colleagues would berate me for that comment because very high bandwidth is actually what you don't want. You want high enough bandwidth and high gain. Finally, optical drivers. These are voltage drivers that drive lasers or modulators, preparing that electrical signal so it can be converted to light through a laser on through a modulator. That's a function, the laser driver function, that we've seen emerge from being integrated within the retimer to now a standalone function because line rates have increased and the signal integrity challenges have become too significant. It's very difficult to use an integrated driver now and increasingly so as line rates increase. In all of this, signal integrity is like a foundational requirement upon which we really have to get creative. My design colleagues have to get really creative. Ever-evolving architectures that perform each of these three functions and push the limits with every generation. Use copper when you can, use optics when you need to. Our view of the data center is completely aligned with that. This is how we think about the portfolio. Scale in or scale up versus line rate. The boundary that you see here between copper and optical may shift, especially as line rates increase. And that's AOK by us because for us, optical or copper, it's all signal integrity. What we're really passionate about is having the best solutions, whether it's linear re-drivers for copper, or TIAs for optical, or lasers and photodiodes for anything optical from scale up to scale out. That's what we love to do, that's all signal integrity to us, and that's our mission. Let's take a look at our CopperEdge product line and why we believe we have the right to win. the industry didn't believe in re drivers we partnered with a major AI supplier and proved that technology at 224 gigabits we proved that we could add 20 DBs of link budget to an ASIC at 224 gig and we're shipping that solution in volume today but what we're really excited about is our 336 gig re driver And here, we're very proud to share a recent proof point. We just learned that our part is the only solution out of six competitive re-drivers that currently performs in an AI data center application. How did we get there? 336 is an emerging line rate. It's bleeding edge. So our engineers worked very closely with our foundry partner on the technology. with the customers engineers to understand the high-speed system trade-offs and how we could push the limits of the process technology in order to get the best possible performance at 336 gig it's easier said than done it took lots of high-speed simulations the requisite whiteboarding of concepts but of course you can imagine that's what you always strive for is getting that dynamic getting engineers in a room together around a whiteboard whenever whenever that happens we know we're going to end up pushing the limits of the technology we're going to agree the right trade-offs where we need extra performance where we can pull from where we can push to and ultimately deliver this kind of success, like 336 gig re-driver. We're now working on other opportunities using the same approach, same hands-on approach, for line card applications, active backplanes, and ACCs. And the market is confirming the thesis. As line rates increase, AI data centers need linear signal integrity. Our fiber edge product line, TIAs and drivers. As I mentioned at the start, we have decades of experience in designing TIAs. From our first two and a half gig products to 20 years later 200 gig. But what does that really mean for us? I want to draw your attention to the the first box here. number one in 100 gig per lane TIA's that wasn't our first hundred gig part but it was our best and our customers have confirmed time and again that it's the best part in the industry and that's actually not new for us we did the same at two and a half gig we did the same at 10 gig we did the same at 4 by 25 gig and we're going to do the same at 200 gig oh and for us to take our market share from 18% to 15% the only incentive we had to use was the performance of the part at 200 gig our shipments have only just begun but our engineers are super excited about what they're developing we're super excited about what they're developing and they're really eager to surpass that 100 gig record and that 200 gig solution together with our 200 gig laser drivers is going to be the foundation going forward initially in 1.16 modules but then quickly in other form factors NPO XPO and other highly integrated form factors of course we're looking forward to that challenge and the ensuing success let's talk about Semtech Photonics, our latest offering. As you can imagine, as TIA and driver designers, we've spent decades designing the parts to work with the available optics. Now we actually get to work with the optics designers in-house. We can both optimize our designs, we can deliver the highest performance, the most efficient solution, end-to-end. Three reasons why our photonics portfolio wins. number one power efficiency 44% power conversion efficiency at 25 degrees less than 0.8 DB of coupling loss versus about 1.2 to 1.5 that's typical in the industry today that means we save 40 to 50 milliwatts of power in an 800 gig module number two signal quality the lasers have very low rim very tight line width. They're essentially coherent light ready today. And number three, we're a US supplier of indium phosphide. With HIFO, we have 46 years of photonics chip experience. We're vertically integrated from epi to chip-on submount. And with DTEC, we're broadening the portfolios to include photodiodes too. And of course, all of this fits perfectly in our scale-up, scale-out, scale-across portfolio. Delivering performance and quality at scale. I'm sure you'll all agree that's the order of the day. We've participated in many very significant build-outs throughout that 24-year history of ours. The initial hyperscaler deployments in the data centers in the 2010s, the ultra-high-volume fiber-to-the-home build-out, the very high-volume build-out of 4G cellular. We've been delivering high performance at scale for 20-plus years. It's part and parcel of our long-standing relationship with our fab partner, Tower. you've all seen the press release earlier this week that was just the icing on the cake our supply agreement with tower supports our financial plan all the way out to beyond FY30 we were actually towers first optical customer for their silicon germanium process all the way back in 2002 and we've been a lead customer for tower since then our engineers have been collaborating ever since. They've been pushing silicon germanium to the next level of performance for seven generations of process technology, driving up bandwidth and driving down power, all the way from 20 picojoules per root hertz to one picojoule per root hertz, and even lower. We began at two and a half gig, and now we're deeply engaged in pushing to 400 gig. So that's the underlying foundation of the technology and the innovation but how do we leverage that for our customers whether we're working with a optical module designer or a hyperscaler engineer we're always striving for that example that I gave you earlier that same approach that we have that we applied at 336 gig in the linear e-driver development getting the engineers around the table getting them around a whiteboard. You can't push the limits of technology by throwing a solution over the wall. It doesn't work like that. Whenever we work together at the engineering level, that's when we know that we can really change the world. Let's look at the various form factors that are emerging for 1.6T and beyond. We call ourselves form factor agnostic. Whether it's copper, FRO, LPO, LRO, employing our linear read drivers or our TIAs or our drivers, these are all opportunities for us. Let's look at highly integrated or look at the evolution of integration, should I say, from LPO through NPO to CPO. LPO, the module sits on the PCB, is plugged into the into the PCB, and the ASIC sits at some distance. In the NPO application, now you have the addition of an optical engine that sits on top of an EIC, an electronics IC, that would contain our laser drivers, our TIAs, and perhaps even our linear re-drivers. And ultimately, the objective there is to increase the density of that module. So today, you're all very familiar with a 1.6T module, an OSFP module, and NPO would nominally instantiate two of those, but even up to four of those. And CPO would, of course, go even beyond that. Let's look at an active copper cable. This is one end of an active copper cable. That copper cable has multiple strands that you can see here are terminated in the PCB, and that module can plug in to a rack in a data center. Active backplane. I mentioned this briefly. Here you can see our linear re-driver chips sat on the backplane itself in order to provide that tailwind, that boost that I talked about. traditional optical module as is being shipped today OSFP 1.6 T employing our various components. An example of an NPO 3.2 T you can see the scale it's probably about that big it's it's very compact fully linear low power high density in order to work its way into scale up in the data center. And then XPO, which is a very high density module form factor, but I'm not going to explain it to you. We talk about partnering with our customers. We have here today Sunil Priyadarshi from Arista, Senior Director, the force behind XPO. So I'd like you all to welcome Sunil to the stage. He'll join us, and he'll talk about XPO. Sunil, good to see you.

Sunil Priyadarshi, Other

Thank you, Imran. Hi. Thank you, everybody. I'm very thankful to Hongho for inviting me. I want to talk a lot about why we need the higher bandwidth for the AI. Panel was very clear about that. Actually, I'm holding this is, most of you are aware, is the 1.60. This is the most successful MSA in the optics industry. Even this year, there will be 100 plus million of this optics, 800G 1.60 will be made. Next year, that quantity could go 200 million. So there is a vibrant ecosystem. There are TIAs, driver, lasers, PDs, DSPs, everything. And there are 100 plus vendors that make these optics. So what is the problem? Why we cannot keep using for the AI? What is the problem? So problem is very simple. Density, bandwidth, latency, and power. Let me give you the example. If you want to connect the 1024 GPU, if you assume each GPU has the 25T of bandwidth needed, if you're using these optics, you will need the 16 racks. So now you can think about you have the four racks of the GPU. To connect them, you need the 16 optics rack. So that make your footprint pretty big. Your interconnect length will be longer, and that gives you the latency, higher power. Probably copper won't work. You need the more expensive interconnect. So how do we deal with this thing? But if you can get something for the AI application, which is very close to this, because this is pluggable, very good ecosystem. easy to build. Actually, I have seen so many videos, automated factories, these are built now state of the art. So why don't we get the optics which is more close to this, not like innovation could be very painful, but time to the market is what we need because So the 200T switch is just coming next year. So we need the optics which can meet, give the highest density if you go with the 200T. So that's what is the XPO, okay? It's an equivalent of the 8 of this. So bandwidth is 12.8T. So now let's do the math again. So if you use this XPO, again, for the same, like, 1024 GPU to connect, you just need four racks. So the reduction in the footprint is like 16 versus the 4. And so XPO can give you the 200T. You can get into one OU. If you want to have the 200T switch with the OSFP, you need the 4 OU. This gives you the 1 OU. If you look at the rack level, this will give you a full rack. It will give you 1.6 beta. This will give you 6.5. And that's what you need for the AI. So what are the other things for the AI? It gives you the bandwidth. It gives you the density. Let's talk about the power, okay? So XPO, actually, we have proven already. It's worked with a full retimed, half retimed, and we also recently presented some data. If you use XPO in the switch, you can do the linear. So XPO does meet the linear on the top. XPO uses the same ecosystem. That is the reason that XPO MSA, we got 175 vendors joining the MSA. This is the highest any MSA, like the company joined a single MSA. Why? Because people are already building this one. They know how to build it. XPO also, they can build in the same time. They don't need a lot of innovation time. That is already existing technology. It made the density same like the CPU. It gives you the wider ecosystem. ecosystem, then CPO, you are just tied with a few like the very small ecosystem. That is the reason like the XPO is like getting a lot of momentum and it needs solve the optics for the AI now and like XPO can also go to 400G. then this form factor can give you 25T. So now think about why optics bandwidth is needed so large because now you think about the 1,000 GPU to 100K GPU. So in the future, like the GPU bandwidth will be going to increase, is now it is 25T, future it could be 100T. So the bandwidth needed is not linear to the number of GPU. Like if you were increasing 1K to 100K, you're increasing like 100 times, but future GPU will be also needing more bandwidth, like 100T, 25T to 100T. So that means bandwidth, network bandwidth, has to go 400 times. Correct. because math is simple very simple I think a lot of people talk about a CPU CPU ultimately will happen most probably at 400 G and actually in in the meantime if you can use like best of the pluggable and best of the CPC because XPO is like the NPO, because the main purpose is how you can make the insertion loss lower. And XPO uses the CPC and the flyover cable, so as you know, the flyover cable can have four to five times less insertion loss than the regular PCB. So this is kind of sweet point for the current AI need the optics bandwidth right away so and easy to build and next year you will see we'll have the 200T switch in one OU using the XPO and one of the more thing is that CPU there are a lot of new things to be learned especially putting the fiber inside the box that is the most delicate part of optics. Everybody knows that. But XBO can avoid everything. It has a coaxial cable inside. It gives you the lower insertion loss. It supports the linear channel. And if future XBO can support 25T, then it will give you even 400T in one Thank you. Thank you very much.

Imran Sherazi, Other

Thank you, Sunim. Really appreciate that. Maybe I can maybe I can take this one. Awesome. Thank you. Thanks so much. Please give Sunil another round of applause. And pardon me for one moment. It's hard to believe that that XPO module instantiates eight of these 1.6T modules in a pluggable form factor. That gives you density, it gives you pluggability, it gives you the ability to switch in and switch out if something were to go wrong. Very exciting. And again, for us, it doesn't matter what the form factor is. For us, the challenge remains the same. And we're extremely excited about it. Let's look at a form factor like this and how it fits in an AI data center. So what you're looking at there on the left-hand side is an optical module. I'll put it the right way around, like this. And what you can see inside that optical module are our products. So two TIAs, four channels each at the top there with photodiodes, eight of them. A driver IC, a laser driver IC, eight channels, and a CW laser source. That's the optical module. The active copper cable looks very similar. I would encourage you all to participate in the show and tell at the back of the room after lunch. You'll get to see active copper cables and optical modules like this, but inside an active copper cable, as I mentioned earlier, the various channels are implemented as single strands of copper terminated on the PCB, and then our linear re-driver chips, our copper edge chips sit on that PCB. But where do these go? How do these fit inside a data center? Let's have a look at that. These repeat across the data center. So there you see the optical module and our ICs fit into that as I just described. But just as an illustration, 20 of those modules would fit inside a compute tray about 16 inches wide those compute trays would sit in a rack eight of them in in one rack a little bit taller than myself so 20 modules in a tray eight of those trays in a rack and 64 of those racks in a cluster it's like a corridor of racks actually in the data center they're on both sides you're surrounded and there are 49 of those clusters, at least in this illustrative example that we're providing, 49 of those clusters in a data center. So we start with two of our TIAs, a laser driver, CW source, photodiodes, 20 of them in a module, 20 of those modules in a compute tray, eight of those trays in a rack, 64 of those racks down a row and 49 of those rows effectively and the net result is 500 000 modules in in this example of a data center it's a staggering opportunity the whole point of the data center is to move data but it's signal integrity that enables the movement of that data and as i've said a few times especially as the line rates go up it's a generational opportunity for us that's how we view it it's everything we've been doing to date culminates in in this so I'm confident that by now you've appreciated that we have both the experience and the expertise to deliver success But what does that mean for you in terms of dollars? We expect to more than double our data center revenue this year and more than double again next year. That's what we call a generational opportunity. And that's a generational opportunity that we've been able to realize as a team. and it's a very fitting milestone for such a passionate team that's been winning together for 24 years. And honestly, it feels like we're just getting started. And I really encourage you to meet some of the team members at the back of the room here to feel that passion and energy that they have for achieving these targets. Everything I've just talked about from copper edge re-time when you must and re-drive when you can to the design team's excitement around what we're building for 200 gig to the photonics opportunities that are just getting underway and our content opportunity in the optical module growing up to 10x we're literally having the ride of our lives and this is giving us great line of sight to our FY30 financial targets. We expect to outgrow the SAM, all these new opportunities that we're engaged with, all these solutions that we're making happen, extending the reach of COPPER, participating in all the different module form factors, developing the best solutions for all of those form factors from a common base of to start with 200 gig TIAs and drivers, getting the right footprint, getting the right pitch to proliferate into all those different module form factors. And all of that's giving us great line of sight, allowing us to outgrow the SAM ultimately to get to 2.1 billion dollars by FY30. So some key takeaways. Signal integrity is what we've been doing for decades and the AI data center opportunity it's not the first huge market inflection we've succeeded in but it's definitely the biggest and what's even more exciting this time is the opportunity is not just optical. It's both copper and optical, and we're just as passionate about driving both of those solutions for both of those. And again, the whole point of the data center is to move data, and it's signal integrity that enables that movement of data at ever-increasing speeds. It's a generational opportunity. And so watch this space. Enjoy the ride. Thank you. And I look forward to meeting many of you over lunch and at the demo stations after lunch. Thank you.

Mitchell J. Haws, Head of Investor Relations

Thank you, Imran. At this point in the program, we will take a break for lunch. Lunch is served across the hall. I would encourage you to echo Imran. There's a great opportunity to see the products in action. So during the break, use the opportunity to do that. We will reconvene, for those of you on the web and in the room, at 1 p.m. Pacific. Thank you.

Operator

Please welcome to the stage, Semtech Executive Vice President and Chief Financial Officer, Mark Lin.

Mark Lin, CFO

Good afternoon, everybody. And just allow me to convey how truly grateful I am for your attendance, both in person and online. I hope you come away today feeling that this was time all spent. You know, for me, seeing Imran share his passion for signal integrity with all of you is truly gratifying. Imran has done plenty of teach-ins for me. His team has done plenty of teach-ins for me. And Ohong said, winning is infectious. Man, to me, signal integrity is infectious. I love this stuff. I also feel privileged to be one of the voices today, representing my Semtech colleagues, to highlight the results of our accumulated efforts, the financial transformation we've been able to achieve together, and our path for accelerating growth. You know, about 10 quarters ago, we were incurring staggering cash interest costs, which resulted in cash flows that were flowing the wrong way. Today, that picture has changed. We're in a position to raise our Q3 27 outlook and provide financial targets, extending to FY30 that shows Semtech's growth trajectory is closer to the beginning and far from the end. So let's get started. My key messages first touch upon a bit of history aligned to Semtech rising and moves to our current state of Semtech transforming and our future state of CEMTECH excelling. First, CEMTECH's financial transformation paved the path for accelerating growth. I believe we first set the company on firm financial footing and were primed to capitalize on market tailwinds and accelerate our growth. Second, I believe we've demonstrated consistent execution in meeting our financial targets. A critical track record as we provide targets through fiscal year 2030. Third, we expect a positive step change in margin profile supported by growth in data center and Laurel. Two very high gross margin products fourth our cash flow generation portfolio transformation and disciplined capital allocation enables organic and strategic investment to strengthen our portfolio value fifth our differentiated portfolio of high performance solutions paired with strong margins and balance sheet stability is expected to compound into revenue, earnings, and cash flow growth, resulting in increased shareholder value. We'll highlight this point through our FY30 financial targets. You know, one of the hallmarks of Semtech Rising was to fix the foundation, enhance our capital structure, enhance employee engagement, enhance customer engagement. What I've concluded is that these three items are remarkably linked and I believe we have meaningfully improved the state of all three. We've de-risked the balance sheet through repayment of high-cost term loans. We've managed debt maturities. We've lowered our cost of capital and we've improved liquidity. About 10 quarters ago in FY24, the year I joined Semtech, net cash interest cost this company eighty two million dollars and free cash flow was a negative one hundred twenty three million dollars clearly unsustainable today net cash interest is a contributor to free cash flow and annual free cash flow as of Q2 of 27 was one hundred ninety three million dollars likewise net debt at the end of one FY 24 was 1.3 billion dollars with a net leverage ratio of ten point five times also clearly unsustainable but at the end of last quarter those figures are down to 299 million dollars and 1.1 times over leverage is in our history but it is history within centex ecosystem including employees suppliers and customers our financial transformation has allowed us to amplify our signal and reduce the noise first in R&D we've been able to accelerate the pace of investment with a laser focus in our high growth areas and generate top-line growth and bottom-line return and certainly better return than we were earning from interest expense I've been able to witness firsthand this multiplier effect on our engineering teams from our increased but very focused and disciplined investment after ECOC last month I made a stop at Semtex location in Switzerland to visit Laura development team our increased investment has energized this team and it's not for the spend it is because this team understands where the market is heading they're listening to the customer and now they're able to accelerate the development of the next generation of LoRa. Our next generation LoRa device, it really focuses on customer requested enhancements, substantially lower standby current, which dramatically extends battery life, increasing data transmission rates, and all with SDKs facilitating device deployment. Second, we've been able to contribute to growing EPS as interest savings have resulted in increased earnings. Third, we've been able to substantially increase assurance to our suppliers and our customers, which I believe is critical in today's accelerated product development timelines. suppliers have increased their credit limits on the customer side I have not had a discussion with a supply assurance team to discuss semtex financial condition in some time that's all been taken care of but more importantly and I believe most importantly during customer meetings our sales and field engineering teams can start a meeting with a customer you know those white boarding sessions that Enron was talking about we can start those meetings discussing and leveraging Semtech's portfolio to assist the customer those teams don't have to go into these customer meetings first having to discuss Semtech's leverage we've demonstrated consistent and predictable execution. Over the last 10 quarters we've met or exceeded our midpoint revenue and EPS outlook, all while growing revenue in each of those quarters. We met these quarterly targets while growing revenue 66%, adjusted EPS 1,083%, adjusted gross margin 470 basis points, and adjusted operating margin 1220 basis points i highlight this track record of execution as i believe it is foundational in giving you confidence in our ability to execute on our multi-year model speaking of execution we're updating our outlook for q3 of 27 to the upside allow me just to review some midpoint figures for this these updated expectations Revenue at $417.5 million, which would be 56% year-over-year revenue growth. Consolidated adjusted gross margin at 59%, and adjusted gross margin excluding the health or cellular modules business at 64%. Consolidated adjusted operating margin at a record 33%. Adjusted diluted earnings per share at a record $1.14, up 138% year-over-year. On the divestiture of the cellular modules business, we continue to expect that the transaction will receive regulatory clearances and that we'll be able to close this transaction in our fourth quarter. On portfolio optimization and moving to a higher margin mix, Growth in our portfolio has shifted dramatically to Data Center and LoRa. In FY24, Data Center and LoRa each comprised about 7% of total revenue. Moving to our expectations for Q3 of 27, excluding the cellular model divestiture, Data Center is expected to climb to 40% of total revenue. LoRa revenue is expected to constitute 19% of total revenue. Translating these percentages into dollars, for all of FY24, data center revenue was $61 million and is now expected to grow to $145 million over a single quarter, Q3 of 27. Lower revenue was also $61 million for all of FY24 and is expected to be $68 million for Q3 of 27. Both portfolios are at record levels and contribute amongst our highest gross margins. And growth expectations through FY30 for data center and LoRa are the basis for our belief that we can grow beyond this quarter's expected 64% adjusted gross margin excluding the cellular modules business. Semtech's margin performance is highlighted here. I believe it reflects a clear inflection. I'll cover historical growth and compare it to our midpoint outlook for Q3. Q2 of 25 to Q2 of 27, over eight quarters. Adjusted gross margin increased 410 basis points. For Q3 of 27, over one quarter, we expect a 950 basis point increase in adjusted gross margin, excluding the cellular modules business. For Q2 of 25 to Q2 of 27, over eight quarters, adjusted operating margin increased 1,020 basis points. For Q3 of 27, over one quarter, we expect an 860 basis point increase. This is translating to adjusted earnings per share. Q2 of 25 had the headwind of interest expense, pretty substantial interest expense, and reported adjusted EPS of 11 cents. For Q2 of 27, adjusted EPS was 71 cents, or growth at 9.3 times sales growth compared to Q2 of 25. For Q3 of 27, adjusted EPS is expected at $1.14, with growth at 2.7 times sales growth over a single quarter we're returning to gross margins that this company has earned before but this time with top-line growth several times higher and operating margins and earnings per share above what Semtech has previously been able to deliver. Scale up, scale out, scale across, across copper, optical, photonics. They've all been contributing to Semtech's scale at the revenue level, providing operating margin, operating leverage, and earnings growth. I'd now like to highlight capital allocation as it relates to operating expense spend. The first point is absolute. We focus investment in R&D and allocate resources to high return projects where we have customer alignment. I think this is evident in the 27% increase in R&D dollar spend between FY25 and the last four months ended Q2 of 27. Our R&D opportunities are vast but we focus on customer alignments. Our ROI calculations factor where customers have invested resources on the project we are supporting. We believe these factors result in greater assurance in the quality of our ROI expectations. Even with these opportunities though, we do maintain an envelope of affordability and have roughly maintained R&D spend as a percentage of revenue over the periods we're presenting here. One way to afford R&D is to watch the efficiency of SG&A and especially G&A. While R&D investment increased 27% on this chart, SG&A increased 8% or 30% of that rate over the same period. We are investing in our commercial team. Listening to our customer is important but you can see as percentage of revenue. SG&A declined from 18% to 15%. Scale is another factor in affordability. Even with increased R&D investment, our Q3 27 outlook pegs the midpoint adjusted operating expense as a percentage of revenue at 26%. That would be down 910 basis points from FY25, down 670 basis points from FY26 and down 600 basis points from the trailing 12 months ended Q2 of 27. Turning to the balance sheet, we ended Q2 of 27 with $503 million of debt against $204 million of cash and cash equivalents for net debt of $299 million. Combined with $357 million of availability on our line of credit, it. That gives us total available liquidity of $561 million. Our debt maturity schedule is clean. $100.5 million comes due in FY28 with no scheduled maturities in FY29 or FY30. And the remaining $402.5 million isn't due until FY31. With expected Q3 trailing 12-month adjusted EBITDA of roughly 360 million dollars and growing cash generation our net leverage ratio is expected to move under one times and toward our maintenance net leverage target of less than 0.0 times we believe low leverage and ample liquidity gives us significant flexibility to drive long-term shareholder value through disciplined capital allocation this brings me to our go forward capital allocation priorities first organic investment customer aligned R&D focused on our highest return opportunities with the clearest visibility second portfolio optimization applying a discipline framework for opportunistic capability driven M&A prioritizing strategic fit and the strongest returns, and staying mindful of our history. Third, balance sheet strength, preserving the flexibility to act on operational and strategic opportunities as they arise. Let me ground this in a recent example, our acquisition of Semtech Photonics Products, or HIFO, in March of this year. Semtech Photonics Products, or SPP, expands our AI connectivity portfolio into photonics and adds domestic indium phosphide manufacturing expertise that would have taken us far longer and would have been far more costly to develop organically. The economics are attractive. SPP's photonics portfolio is expected to generate data center level gross margins with clear R&D and go-to-market synergies. And the transaction was appropriately sized to preserve our balance sheet strength with capacity expansion both what we've spent and what we expect to spend within our liquidity framework this is the model going forward opportunistic capability driven acquisitions that strengthen our core franchises not scale for its own sake so allow me to translate all of today's presentations into our long-term financial framework. Our accumulated efforts in revamping customer engagement, the concerted R&D investment in high-return customer-aligned projects, our continuous portfolio optimization, the multiplier effect from stronger employee engagement, and assurance from our financial transformation together form the foundation of the framework we are calling Semtech excelling. We expect revenue to grow at a 38% CAGR from FY26 to FY30, reaching $3.2 billion. I want to prove precise in the base and that we exclude $166 million of FY26 cellular module revenue, the business that we expect to invest in Q4. Data center revenue is the compound at 75% to $2.1 million for FY30. As Imran mentioned, we also expect data center revenue to more than double in FY27 versus FY26, and then more than double again in FY28. Imran showed you why. Fiber edges differentiated performance. Copper edges differentiated performance. photonics with differentiated performance just starting to contribute to revenue content opportunity per module growing at over 10 times a strategic supplier with whom we've had a long-term supply relationship aligned to supporting multiple definitional customers and all across the sam that we expect to grow from three billion dollars to $15 billion. Laura's revenue is expected to compound at 30% through FY30, supported by capability enhancements and expanding market verticals, including smart building, security, drone control, edge AI connectivity, and our emerging partnership with Amazon Sidewalk. We expect steady growth from the rest of portfolio above market growth rates and contributing cash flow and contributing operating leverage. Alongside revenue growth, we expect adjusted gross margin of 70% in FY30. For this quarter, excluding the cellular modules business, we expect gross margin of 64%. That remaining six points are contributed through mix with data center and lower growth and growing scale of our business. Adjusted operating margin for FY30 is expected to range from 48% to 50% with R&D investment as a primary variable. And of course, incremental R&D would follow our same framework of customer-aligned projects. We also expect our growth to result in healthy cash generation, with free cash flow as a percentage of revenue to be 35% in FY30. Together, that drives adjusted diluted EPS CAGR of 64% to 66% or adjusted EPS at $12.40 to $13. We've assumed a 20% non-gap tax rate based on geographic mix of income and 104 million weighted average diluted shares outstanding. To close, this is our model to compound shareholder value. Winning in high growth markets by aligning investment in our customers needs increasing portfolio value by building a more differentiated higher margin portfolio expanding earnings power by turning growth into stronger earnings generating durable cash flow by converting those earnings into durable and sustainable free cash flow and then reinvesting with discipline to deploy capital to maximize returns quality top line growth margin expansion cash generation disciplined reinvestment that's how we intend to compound shareholder value in the years

Mitchell J. Haws, Head of Investor Relations

ahead thank you please welcome back to the stage mitch haas thank you mark at this point we will open the floor to questions. I'd like to invite members of the senior leadership team to join me. Just a couple rules of the road. There are mics in the audience. Before you pose a question, please use the mic for the benefit of those on the webcast. And if you could, please state your name and the firm you're associated with. With that, we do have two additional members of the team that haven't presented today. So I thought it might be good if they just made a quick introduction. Jason?

Jason Green, Other

Sure.

Mitchell J. Haws, Head of Investor Relations

Jason Green. I'm the chief commercial officer, so I lead the customer-facing teams and the go-to-market effort.

Imran Sherazi, Other

Esaf? Hi, everyone.

Hong Hou, CEO

My name is Esaf Silberstein, and I'm the chief operating officer in the company.

Mitchell J. Haws, Head of Investor Relations

Okay? So the first question. Chris, fire away. Chris, can you excuse me? Before we start the Q&A session, I have a very important announcement.

Hong Hou, CEO

As we were preparing for the analyst day and Mitch was preparing to be a grandfather. Great day today, Mitch became grandfather.