Hi, good day. Thanks for joining us. I am Mike Genovese from Rosenblatt Securities, the Communications and Cloud Equipment Analyst. This is the Rosenblatt Age of AI Tech Conference. Super happy today to be joined by VIAVI, one of my favorite companies, ticker VIAV, we have the CEO, Oleg, and the CFO, Ilan, for you. Welcome, gentlemen. Nice to see you. Oleg, Thanks. Just to frame things for people at a very high level. The company has really changed over the last couple of years. It seems like it happened all at once, I think that it was the result of deliberate positioning to go from a really over-indexed telecom company to one with a lot of exposure in the AI and data center world, as well as a really interesting mil-aero business. Just help us frame that transition and managing what you saw coming, what you did at a high level to get us where we are today from where we were just a couple of years ago. Sure. Thank you, Mike, it's a pleasure to be here. I joined this company 10 years ago, just as JDS Uniphase split into Lumentum and VIAVI. At that time, we had two businesses, network test and measurement, optical business. Networking test and measurement was very heavily, almost over 90% service provider-centric, selling field instruments for a lot of copper testing, DSL, cable, of course, fiber. When I joined, we clearly saw that there was a number of big inexorable trends. One is copper going away, we had to really just redouble on fiber. Also, telecom service provider is not exactly the best ZIP code or neighborhood to be. It's very volatile. The spend profile changes a lot with the health of the operators. The first thing is I said, Okay, we need to diversify. We cannot be in the ZIP code. The most easiest thing is you take your engineering knowhow, your capabilities of just communication, a lot of fiber, said, You know what? In stead of field, we'll continue maintain investment in field, we really need to get into a more lucrative test and measurement market, which is lab and production. Sure. We're not going to do copper, we're going to do all fiber. That meant taking a lot of these capabilities and developing bench-top instruments for optical testing, which is targeting semiconductor vendors, module vendors, system vendors, and also production equipment, also targeting optical modules and system vendors. That was put in place about nine years ago. Over the, I would say nine years, we are now in the third generation of these products. We also have clearly done the communication as well as PCI Express, and of course, when the data center market took off, there was additional bonus to it because you went from technology being replaced every 6-8 years, driven by service providers, to being effectively turned over every two and a half years, driven by data centers. That growth and acceleration of product has really drove the business quite significantly, and that business has been growing for us by leaps and bounds over the last two years. At the same time, about six years ago, we also said, We're missing wireless. There was a good opportunity to acquire wireless test measurement business from Cobham. Through that, we inadvertently got into aerospace and defense. As we looked at the business said, "That's a very attractive ZIP code, but we don't want to compete against those big 800-pound gorillas. Let's think about it. How do we intercept it? What's the future?" The future we saw was assume that GPS is dead, and you have to operate in the GPS-denied environment, and what would you need? We found that that space was extremely underserved. By focusing on this whole area called PNT, positioning, navigation, and timing, we made a number of smart acquisitions, as well as internal development. Of course, now we all see the war of the drones. That has now completely flipped the whole narrative on defense environment that you have to operate in a GPS-denied environment. In addition to having a massive growth in the AI data center ecosystem, we also have a very strong growth driver in the aerospace and defense because what we do is pretty much everything that requires you need to have to operate in autonomous mode, independent of GPS. Yeah. So- Hey, Mike. Before we continue, sorry, just to read the obvious safe harbor statement because we haven't started recording. Okay. Obviously, many aspects that we'll be discussing today and talk about today are forward-looking in nature and are subject to risks and uncertainties. These are valid only as of today, and we don't take any obligation to update these statements. Thank you, Ilan. This statement for a rearward-looking, but thanks for the update on it. Sure. Great. Perfect. Oleg, I want to double-click on that data center business and that mil-aero business, but before we do that, because I really think about those being the growth drivers of the business, and I'm almost modeling everything else flat. Is that too dour on everything else? Ciena just reported 28% growth in their telco business for the quarter, which was higher than we expected. Is there any signs of life, and should we be modeling the rest of the business to actually be a growth business, or really just thinking about those two segments that you talked about driving? No, that's a good point. I think the service provider business, it's growing in a low single digits. That's nature of the beast. Also remember there's two story of two service providers. There's your traditional service providers. These are the guys growing at low single digits. There's the next gen service providers who we actually put into the data center bucket. These are companies like Lumen, Zayo, and others who are doing the whole interconnect between all the scale across and scale out. We actually treat it not as a service provider. We treat it as part of the data center ecosystem because all their requirements, their service level agreements, and their spend is driven by data center operators. The traditional, your classical service provider, wireless carriers, that's a low single digit. The wireless business is still pretty slow. Even that one it's bottomed out and starting to grow in the low single- digits. I think we probably will see acceleration in the wireless service providers in the next year or two as inference moves towards the edge. I imagine there'll be money coming from outside because I don't see the incumbents really having wherewithal and the financial muscle to actually take advantage of inference at the edge. They own the spectrum, so there'll have to be some kind of partnerships. On the business, I would say it's a mid-single digit growth. Have you looked at quantifying the, when you talk about AI moving to the edge, the size of that for somebody like Nokia or Ericsson and somebody like you versus if there was actually investment going into upgrading 5G capacity and 6G capacity? Is it a smaller opportunity or is it- So- Could be that big it should be huge. From all the things I've seen it, the inference market is supposed to be around 3x of the training market. I would say we're not even talking about first inning, we're talking about pre-game show. There's talk, you saw NVIDIA gave Nokia $1 billion to start developing AI RAN. The radio access network is woefully unprepared. Once they will build it and they say, Well, now we need to monetize it, there will be a rude awakening, just like it was with the fiber, that the network is not ready. When the big four or five realized the fiber was not ready, they started investing aggressively into it. We've seen since then this level of interconnect fiber really go up significantly, which has been extremely good for VIAVI. I see the same thing happening on the edge. You have today traditional service providers who own the spectrum, and they own a RAN, and even that RAN is only maybe 30% 5G built out. As we go to AI RAN, it needs to be boosted up to all 5G, and then on top of it, you need to improve the latency and fiber performance. I expect that will be the next wave of spend that we will be seeing. I would imagine it'll be money coming from outside because I just don't see the incumbents having the wherewithal and the financial muscle to do it themselves. Great. Let's go back to the data center business now, because I think there's a lot of different things going in the data center, we think before OCS and CPO become larger and probably really large. We think about the business being driven by transceivers, the transceiver market looks like it's doubling this year. Next year could be anywhere from up another 60%-100%. As we track the transceiver market, if we see upside to our expectations, should we then just automatically assume that that helps VIAVI? How do we think about the growth of the transceiver market, and then how that flows to VIAVI? I would say if you think about the transceiver market, think of the second derivative of growth. If you double the amount of market, means you need to double the manufacturing capacity. We grow purely with the size of the market, that market's been growing like crazy. Just to give you the products we sell into transceiver test, is more than tripled in the last two years for us, it continues to increase. It's going to get even bigger. Let me just tell you. The first thing it drives you is the tide of annual demand. As you put in capacity to support a certain amount of transceiver market, it scales linearly with what we need to ship. If you double the transceiver market, we need to double the installed capacity. If you add 100 million units, let's say 10 million units, whatever, it's purely with the second derivative, which is the incremental amount of installed capacity. That's one thing. The second thing is. As you go from 400- 800, now we're talking about 1.6 Tb modules, 400, 800, a lot of it is copper. There is some optical testing, and that kind of was going linearly. As you go to 1.6 Tb, it's predominantly all optical. Now you have not only you're increasing the capacity, you're increasing the scope of tests. Now you have a two-dimensional growth. You have the growth of the unit volume, and you have the growth of the content. That is like if you double the installed capacity, you actually do more than double of test equipment. The whole new animal that we are seeing, the whole co-packaged optics, we haven't even gotten to that. That is increasing the demand multiple amounts of what I was talking before. As you get into co-packaged optics, you're looking up to four insertions, and that's the business we were not even playing before. Now you have our optical equipment moving not only traditional transceiver or the NEM testing market, you're moving into the, what would be the old OSAT, outsourced packaging and assembly, and test market. You're moving into the semiconductor packaging where you do multiple insertions testing. You would have a prober like FormFactor, you would have a test system like Advantest, for example, and you would have a rack with all the VIAVI equipment that doing all the optical testing. That business for us, I'd say first inning would be beginning in the fall. It's just about to start. I don't know how big that market is going to be, and it'll probably depend on the yields and the speed of adoption of the co-packaged optics and the Photonic Integrated Circuits embedded into substrate or interposer or things like that. Right. For the people that are not familiar, although more and more investors are hearing about this, right? These four insertion points you're talking about in co-packaged optics, I think it's a electrical IC, an optical IC, an engine, and then an ASIC, right? I think those are the four. If we kind of simplify it for people, we can sort of say, well, previously you were testing really focused on the optics, some electrical, but mostly on the optics, and now you can also test the chip. Is that too simple, or is that not? Previously it was done at the final test. You have the whole module, right? You're just in and out, and it's good. This one is a lot more because this goes into these, it's called heterogeneous integrated packages where you have CoWoS chip on wafer on substrate, this whole big thing. My past I used to run Amkor. The problem you have, as nice as those system in the package, despite all those lingo, think of it as a system in a package, but it's a way more complicated system in the package. The problem with system in a package, as good as it gives you the latency, the power, all the benefits, the yield is a killer. Before you put this $1,000 ASICs at the end, you're basically doing additive manufacturing. You're starting with a substrate interposer. You put in the electronic integrated circuit, like the all IC. Well, somebody probes it and tests it. That usually was done before. There is a Photonic Integrated Circuit. Well, you have to test it to make sure that it's good. You put it in, and unlike with electronic integrated circuit, just by putting it and embedding it into interposer or substrate and running it over the heat cycle, you can change the optical characteristics. Now you got to retest it again to make sure now that you put it in, it's still working well. You put the next layers of things on it, and you got to retest it again. At the end, you put an ASIC on, and now you got to test the whole system together. That's when we talk multiple insertions, is just think of it how many times you need to run it through the tester, which is way more than traditional electronic circuits, which is really just been a probe and the final test. Okay. I guess now this, whether you do scale out or scale up, it's the same, right? For you- That's right It's the same. I imagine that when you say starting in the fall, those are the scale out that we're working on now, and scale up comes later. Is that correct? Yes. In turn, you mentioned the CoWoS, you mentioned the interposers. I guess a lot of people are sort of wondering, can TSMC or other big fabs ramp quickly enough for this whole thing to stay on track? Do you have a view on that? Clearly they're spending. If you look at TSMC CapEx, the huge chunk of their spend is on the back end. It's their whole hybrid bonding that they have, the Wafer-level CSP. They can scale. They've been investing a lot of capital, There's also a lot of challenges, like there's a shortage of interposers, right? Because you got to buy silicon wafers. I think also, they're placing massive orders for test equipment, right? It's going to be coming in over the next several years. So they're installing significant capacity. It's not only TSMC. You got other major players, right? It's not only the ASIC and things. Think of all the memory guys who are doing this stack High Bandwidth memories, right? It's the same type of problem, right? You're putting more and more like glass core in a substrate. That's another type of test. We're talking about purely these advanced packages. Let's not forget all the fiber that's going into the data center. There's a whole other, we never used to play in the fiber optic manufacturer. You see these multi-billion dollar agreements that the hyperscalers are signing with the likes of Corning. It's not your grandfather's fiber, it's Multicore Fiber, it's Hollow Core Fiber. It is very advanced stuff that is complete. All of a sudden, what we have is exactly what these markets need. We're seeing, the traditional that you think about VIAVI, okay, how many transceivers there are and what is the second derivative of growth? That's kind of like our old core business. What we are seeing happening on the semiconductor advanced packaging and the fiber, that just creates a force multiplier, and expands our TAM of our market significantly. Great. Recently, people have been talking about more about test and measurement. You kind of hear the bottleneck language, but it's. I want to ask you, is it a bottleneck and can you describe it? Because I think it's different than the indium phosphide where you have to order substrates and machinery and build fabs and order things from you. That's a kind of a capital and lead time bottleneck. When we talk about test and measurement bottlenecks, would you describe it as a bottleneck and could you give us more detail on what you're seeing? I think from our side, at this point it's not, but as a result, people are placing orders for over the next two years and deliveries. I think the, it's like, you ask us how many probers can FormFactor, manufacture anything in time. I think in that particular case, that may be the rate of deliveries. It's not for the shorter, for the semiconductors or. I think in the end, it's like, but also you ask how many systems can be internalized. When they come out, you got to put them on the floor, you got to program, you got to set up. You also got to develop all these different insertion tests. I think it's not, I wouldn't say it's a hardware-driven bottleneck, but there is a, it's combination of all the factors. Great. We didn't really talk about OCS yet, I just heard at a conference today, Lumentum put out the biggest number for OCS I've heard so far as being a $10 billion market. Coherent had said 4+, Lumentum said 10 by 2030. The OCS market keeps getting bigger, what's your exposure there? Well, if that gets bigger, everything I said, just multiply it again by several factors. OCS is the sweetest spot of the optical test market. The more expensive and the more critical is the component, the more tests you have to run. If you have a 300 by 300 OCS, you have to now test every single port to port, every link. Once it goes in, you're switching to 100 Tb or more of data, right? If one of those links is bad, you're losing a terabit of data, just like that. That leads to a much more serious failure than if one of your 400 Gb ports goes bad, right? In that respect, validation, verification, characterization of the switch becomes quite big and, really, if you think about it, am I going to judge what's the right number? I'm not. The reality is, if you can do things in optics, it beats electronics any day because the power budgets and cooling and heating, I mean, powering and cooling, are a fraction of what you need in the electronic. The more you can do in optical domain, I mean, that's if you take a Google's philosophy, the better it is. The technology and science behind optical switching is really has been advancing by leaps and bounds. I do believe you'll see a lot more stuff being done in optical domain, and that's why you see even NVIDIA and you see Marvell making massive investments in the optical space because you still need both electronic and optical, but the more you can offload into the optics, the better it is because otherwise you won't have enough power in the world to power everything. Is it fair from what you're saying, if we want to do analysis where we just sort of took the TAM of transceivers, of OCS, of CPO, and we said, Well, what's the percentage of that market that would be the testing opportunity? It sounds like in CPO and OCS, it's a higher percentage, a richer mix than transceivers. Is that? When you take the transceiver, you pump a laser, you measure the power, and you measure the output, all right? It's in and out, right? It's plain and simple. You could do it for under $50,000, right, per station. When you're talking about OCS and for packaged optics, you're talking about racks with tens of thousands of dollars or $100,000 worth of equipment. That's a very different animal in that respect. As you go from 1.6 Tb and beyond, it's going to be almost all optical. You get actually much richer content of optics than you had, even at 800, because there's so much copper still, you get active cables and all that. We still sell equipment in there because you're doing optical, electric, like EO and O2E conversions. Once you get to all optical domain, then everything you got to do is in the optical domain. When does the OCS opportunity start? It's already started. There's certain hyperscalers been doing their own OCSs for a while, and Lumentum, our cousins, they've been doing optical cross-connects, since the JDS Uniphase days, right? The technology today is much bigger and more complex, this business has always been the holy grail of optical testing. It was mostly for kind of core optical switching and things like that. What used to be something that you did in hundreds of units, now you're looking at thousands, tens of thousands, hundreds of thousands potentially, right? That is really where it's all going. It's not that it's a new technology, and although there are new technologies emerging, you can do MEMS, you can do liquid crystal, there's some of the other technologies. Point is, it all requires optical test, to make sure that all the ports are switching correctly and things like that. What we're seeing now is the complexity of them is going up exponentially and the volumes are going up exponentially. If, again, you got a two-dimensional scaling complexity and volume, and that's what's going to be driving a lot of demand. Right. Earlier when you mentioned that the transceiver-related revenues had tripled, just remind me again what that timeframe was, and also is that include the acquisition or that's not just organic, is that the acquisition as well? Yeah. It's all purely organic, and I would say it's in the last 18 months. Okay, great. I think maybe the last thing on the data center side is just, are you seeing expanded use cases for fiber monitoring by the hyperscalers? Well, that's a very good point. I earlier said a lot of hyperscalers, if I talk to them five, six years ago, which we did, their view of, Oh, fiber. Oh, somebody else will bring the fiber and they'll plug it in, should be no problem. Until they found out the state of the fiber network. A lot of the dark fiber is no longer usable. It's been degraded. They need a very different fiber. They need thousands of bundles coming into the data center. What they're doing now is they're also realizing they need latency, they need performance, and they need characterization. They've been so disappointed with what's out there in terms of the service provider availability, they're taking it into their own hands. What many of them are doing now, they are at the entrance into the data center, they have fiber monitoring that all the dark fiber gets connected into the fiber monitor. They characterize all the fibers. As they increase capacity, and often they switch new fiber, maybe as often as every day a new fiber gets turned on, they know which fibers are good, which ones are bad, and which ones are ugly. They also measuring, the service providers and hold them to the service level agreements on latency and performance. As a result, we're seeing huge demand for fiber monitoring from all these hyperscale AI data centers. The positives, the secondary effect, the service providers, now that they're seeing they're being measured, some of them are saying, Heck, I better put the same thing on my end so I don't get killed on the violation of my service level agreement and get charged penalties. If anything, I would say it's improved tremendously the performance of the fiber network that's being connected to data centers. Also the next level is going to be as you put Hollow Core Fiber and Multicore Fiber, you need to not only monitor latency and performance, you also got to make sure that the quality of the connections being made are adequate, and so you don't have a problem down the road. Fantastic. Good stuff. I guess we should talk about the position navigation and timing a little bit more. Just help us understand why it's so important to the industry, the military aviation industry, and specifically your value proposition in the segment. I would say it's critical to all critical infrastructure, telecom networks, wireless networks, energy pipelines distribution, right? Aviation. Even using your Google Maps on your car, we rely on GPS for timing signal, and that's what keeps all these networks synchronized. If network falls out of synchronization, it shuts down, right? Because you can no longer talk and receive, send and receive at the same time. Think of banking, all the transaction trading, there's a timestamp that needs to be exactly the same on both ends, right? When GPS today is with a $50 device, you can spoof the GPS, you can jam the GPS. GPS is very vulnerable and there's already recognition. There's actually much bigger level of concern that being led on by the regulatory authorities. They just don't want to create panic. If you look at Europe with the war going on between Ukraine and Russia, if you look at the whole Middle East conflict. Commercial aviation cannot function there because the GPS is jammed, so they have to navigate by different beacons and things like that. All right, anything that today as we go more towards autonomous, whether it's aerial, ground, sea, or undersea vehicles, they all need an alternative to GPS. For the last 30 years, GPS was like a holy grail. It's always there. Assume today it's no longer there. There's a whole new industry evolving called altGNSS, alternative GNSS, and PNT, position navigation timing, is at the core of it, right? It's first of all a resilient timing, which requires you be extracting timing information not only from GPS, but from any satellite constellation. Even if that gets jammed, you got to be able to hold over and operate in autonomous mode to provide timing information and then recover. Then, of course, the whole thing, positioning and navigation. In the absence of GPS, you need a lot of inertial navigation sensors and thing we call sensor fusion, where you take multiple signals and infer your position from that so you can navigate and fly or drive or swim. Right. Now, I imagine right now there must be a good amount of drone restocking going on, and that probably is good for your business. Is that fair? It's drones, it's munitions, it's everything, because everything that has smart in it requires these type of products. Great. Well, let's take this to the business model level because everything you've been saying is super bullish. The current growth rate of the company in recent quarters has been about 30% on the top line which is a pretty healthy growth rate. I'm not sure you want to. Well, first of all, let's talk about how far you can see out, because typically VIAVI can only see out one to three quarters, and three would be the best it could possibly get. I think there's probably reluctance on your view to talk about beyond three quarters, given that history, but also given everything you're seeing and those orders you were talking about going out on longer durations. Are you getting at all more confident that you could talk about multi-years or- You're being generous, Mike. The old VIAVI, really dealing with service providers, the most you see is one quarter. Now we have an optical coating business, and there we have an annual forecast. That's a very different animal. I would say where we are today with the data center, I think our visibility is about 2-3 quarters, and it's really built around deliveries. There is clearly business out there, but the delivery schedules don't get formalized. As lead times extend, we're seeing more and more customers providing a longer- term forecast. We tell them, Hey, if you want me to deliver it, I need to place orders for parts today, and I'm not placing it until I have a firm order. I'd say as part of it is, I'd say you get probably today, I'd say 2-3 quarters visibility. Right. CPO starting in the fall, scale-out is definitely happening. Right. We could debate scale-up timing and all of that. It feels to me like the increasing likelihood is if we have three good quarters, we could have three more good quarters, and the next cycle could be right behind this three-quarter cycle. No. This is what I tell everybody is, okay, I think at least I see the next 12-18 months should be just as strong. If you see everything that's already loaded, it's coming out. Of course, then we'll hear what people say next year, how they're going to spend. There's another thing people forgetting. Remember I said the half-life of the data center is about 3-4 years. That means now what you just start doing two years ago, in about 18 months, you're going to look at the hardware you can get, and you said, Hey, within the same footprint and the power footprint, physical footprint and the power footprint, I can do a double, triple, quadruple level of performance. All of a sudden, says, Hey, you know what? My OpEx costs are going to be way lower if I just rip everything out and replace. That's when that cycle starts again. By the way, that's the way all the data centers been always operating. It's always been like a three and a half kind of year replacement cycle. Right. Given what's going on in data center, given what's going on in mil-aero, the rest of the business being a low single-digit business. You put that all together, we are growing double- digits, right? I don't have to ask you if it's a double-digit growth company. Yeah. It is, right? Yeah. I know you don't want to look too far out and give a five-year forecast to things like that, but it seems like this pace that we're on, there's no reason to think it's going to suddenly go away. Is that fair? At this point, I don't see it. In fact, from where we are today and where we're going, I think we're looking at a $2 billion run rate in a not too distant future, right? At least getting to the exit velocity of around $500 million a quarter within the kind of, I'd say four to six quarter outlook. That's the most you've sounded like Lumentum in giving an outlook that I've heard in a long time. If all of their orders materialize. Yeah. I'm being the most conservative of them all because, see, my worry is ultimately The golden screw, right? I can deliver all the things I need to do because for us, we have a pretty good control of our supply chain and we don't need egregious volumes of product, of components. Clearly, we all know memory is going to be on allocation through 2030. If people cannot get memory, or worse yet, they cannot get chip capacitors, that can be a killjoy for a lot of things. Right. Today, everybody's talking about all those big, beautiful modules, which we love, you got to look deeper. I go as deep as looking at the availability of glass core for the substrates. There's only two players in the world doing that stuff, right? Japanese and somebody else. You got to look at all of that capacity, because if they cannot get the substrates, they can't build the products. Ultimately, that's the things. The industry is pretty resilient and a lot of these bottlenecks are getting resolved just by throwing some money at it. Yeah. Far it's been pretty good, pretty resilient. All right. Ilan, maybe you can take a shot at this question first, and maybe Oleg will add as well. On the last conference call where you mentioned that the incremental operating margin, I think you said 40%-45%, which is a good number, right? Since the operating margins now are just breaking 20%. It actually seems if we do the math in our models, it seems like that's a pretty sad bad number, that actually the incremental operating margin might be higher than that. I guess my question is could the operating margin end up, or incremental margin, end up being higher than 40%-45%? The 40%-45% that you mentioned, Michael, obviously is the fall-through that we have been discussing in the last two quarters. We've seen since last year, in the mid-30s, going into the 40s. Obviously, as Oleg mentioned, this growth, we believe will continue, at least based on the visibility that we see so far. As we continue to scale the top line, the fall-through can improve. When we get to the $2 billion run rate, when Oleg mentioned it earlier, yet that 40% can be higher than that, and operating margin can get to the mid-20s or so. Definitely, there is more opportunity with the growth that we see. How about the NOLs? Just remind us how the NOLs impact EPS. Oleg mentioned also earlier our cousin, Lumentum, with the JDSU. The split was that we actually kept all the NOLs, which amounts for about today, about $6 billion of NOLs, of which about $4 billion we converted to amortizable assets. It means that we have way more flexibility to utilize those NOLs. It's also usable for us for at least the next 5-15 years in different tranches. There is obviously a lot of profits that we can offset with those NOLs. If you think about just the recent equity that we raised, relative to the interest that we paid, the high interest that we paid on the Term Loan B, that interest was not tax shielded because we do not pay any taxes domestically. Net-net, it was accretive from day one, since there was no tax benefit for us on that interest. Moving forward, we'll continue to enjoy those NOLs. We try to divert as much profits as we can globally to the U.S. in order to leverage those NOLs. If you look at the tax rates, for us, the effective tax rate was in the mid-20s, then the low 20s. Now it's in the low teens. The more we grow the top line and profitability, the more we'll enjoy those NOLs, and the tax rate will continue to go lower. Great. You mentioned the deal there. Backing up on the deal, what was the impetus for the public offering? A combination of continued top-line growth, continued profit margins that will continue to grow. That's obviously the top priority. Then stronger balance sheet. That's a priority that we have been discussing for the last several quarters. Earlier this year, we said we plan to pay off about $100 million out of that expensive debt that we had. We were able to pay off about $150 million out of the $600 million, and we had an opportunity, with net accretion result, to pay off the entire debt. Right now, I think our balance sheet is much stronger. I think we get more credit also from the rating agencies, and that's part of our capital allocation model to continue to perform. Great. Sounds good. I guess we're right on one minute to go. I guess, Oleg, just any last message for the investors that are listening? The one thing you'd want them to be left with when they think about the stock and the opportunity. Yeah. Closing remarks? I think, clearly we are not just one, but I'd say two very strong growth engines in the company. The whole data center business, which today is approaching 50% of our networking business, aerospace and defense, which is about 15%. About 65% of our networking revenue is growing at double-digit rate. The rest of the business service provider, it continues a gradual recovery. It's a positive contributor. I think there's several other growth catalysts out there, especially the wireless, if and when it eventually Well, it's not if. When it eventually returns, it will be yet another stage that will light up in terms of driving the growth. The nice thing about growth with VIAVI, you talk about operating leverage. When most companies talk about operating leverage, they said, Well, if my revenue grows, my OPEX scales very nicely, I get expanded operating margin. For us, it's more than just operating leverage. It's also the EPS leverage because as Ilan said it, all that incremental profit and the way we structure all our acquisition and our legal entities, we capture as much profit as possible in the United States, all of that is a tax-free for us. We have about $6 billion of NOL. As a result, we have double operating leverage, not only from the top-line growth to the operating income, but also between the operating income and the EPS. Perfect. Really enjoyed our conversation, Oleg and Ilan. Thanks so much for doing this, look forward to following up again soon. My final message to investors is, I hope by viewing this or watching the replay or reading the transcript later, you take away what a strong and seasoned and thoughtful management team is leading this company. Thank you very much. Thank you, Mike. Thank you, everybody. Bye. Bye.