Nokia Oyj (NOKIA) Earnings Call Transcript

Thanks a lot for coming. I know it's hard after one day. I'm sure everybody was at the bar last night. I know when I was sitting in the lobby, people were trickling in, then they were going to the view, they were coming down, it's just -- so thanks for coming. I appreciate you coming, a lot of you investing the time with us to learn about the Nokia, especially with the combination with Infinera, what our plans are. So our intention today is to talk a little bit about the trends we see in the market, the challenges the optical industry is facing and the requirements that are being placed on it going forward, especially in this -- the context of AI. And then we want to talk a little bit about how our optical business, in combination with Infinera, is looking at addressing this opportunity with the expanded scale we have from -- not just from a business volume perspective, but from an R&D, from an expertise, from a reach, from a solution set perspective, how we are coming together really, really quickly, to be honest. And that's quite exciting, to come together and address this opportunity. We have a great lineup of speakers. We have Federico Guillen, our Head of the Network Infrastructure Business, President of the Network Infrastructure business; James Watt, who runs our optical business. Really senior leaders within the optical business, Ed Englehart and Ron Johnson. So they'll all be talking. We'll have a presentation for about 40 minutes. We're going to leave a lot of time for Q&A at the end. But one thing I'm supposed to tell you before we get into that is to read the safe harbor statement, thanks to our IR friends who insisted I must read this verbatim or the lawyers will be -- will put me in trouble. So bear with me, I'm not going to read this whole thing, but I have it condensed, yet 3 paragraphs worth of description. During this event, we will be making forward-looking statements regarding our future business and these statements are predictions that involve risks and uncertainties. Actual results may, therefore, differ materially from the results we currently expect. Factors that cause such differences can be both external as well as internal operating factors. We have identified such risks in the Risk Factors section of our 2024 annual report on Form 20-F, which is available on our IR website. We also remind you that we are currently in a quiet period leading to Q1 earnings and will therefore not be addressing any financial questions or giving any forward-looking statements or commentary or update around the progress during the quarter. So please refrain from asking questions about that. Thank you. So without much ado, I'll introduce Federico to the stage.

Hello, and welcome, everybody. So it's 8:30, and I already broke 2 rules. I was telling Ron that there are 2 rules for speakers on stage. One is don't go to the bathroom after you are mic. And the second rule is never drink coffee before going on stage and my cup is there. But okay, I'll do my best. So as I said, welcome, everybody. As you know, this industry that we are in, in the recent years was kind of a roller coaster. We had a tremendous growth for investments on the COVID times. What happened during COVID, you all know, I mean, we were all locked down. And very soon, you realize that you could even tell who was connected on ADSL, who was connected on 4G, who was connected on fiber. And then the demand on fiber grew. And right after that, the demand on IP and optical. So there was a wave of growth. Then after that, there was a sharp decline. And the decline was because, obviously, all the customers wanted to cover, cope with the traffic growth and they overinvested, not overinvested, but they created inventory that then we had to digest and it came the decline. And then what we are seeing now is it's coming in the perfect moment after the combination with Infinera is a new investment wave, driven mainly by AI and data center and LLMs and all the traffic growth that is coming from that. So -- the thing is that this is the lifeblood of the network is AI. AI, I used to say when we're talking about the metaverse, I used to say, I don't know when the metaverse is going to come. The only thing I know is that I'm going to have to deliver a lot of traffic. With AI, it's different because there is not so much hype, it is a reality. It is there, and it's already generating a lot of traffic that we have to deliver from fixed, from IP and from optical networks. So we are there to support. No wonder, OSC is the place to be in this spring because this is the right place to discuss and try to solve those problems that the industry is going to face. Now our sector, as I was saying before, this has been in a low tide. And Warren Buffett once said, is when the tide goes out, you can see who has been swimming naked. And when the tide goes back up, you see who's got the better boat and the better crew. And at Nokia, we are convinced that now we have the better boat and the better crew to navigate this wave. So in NI, Network Infrastructure, we have 3 business divisions: Fixed Networks, IP Networks and Optical networks. Fixed Networks has been experiencing massive rollouts during the pandemic, as I said. And the key in fixed networks is to cope with that demand that is coming. Half of the households worldwide have fiber coverage, which means that the bottle is half empty because there is the other half that we still need to cover. On top of that, there is a lot of pressure to increase the capacity of the fiber. Once the fiber is there, fiber is the limiting factor of fixed networks, deploying the fiber. I mean it could take 5, 7 years to give enough coverage in a given country. But once the fiber is there, it's a matter of upgrade the fiber. We started with GPON. Then we are now seeing the wave of XGS-PON. And in Nokia, we're offering already commercially XGS-PON and 25 gig PON. And we have tested already 50-gig PON and even 100 gig PON, 50-gig PON with Google, and we have demos on 100 gig PON. Now everybody likes to talk about more capacity, but the reality of the fixed networks war is it is not so much how much capacity can you deliver as how much money the end user is willing to pay for a given service. So many times, we have seen that we have a technology that is much better today, but you need to wait a little bit until the cost points match the end user demand in terms of cost. So we are getting ready into 25-gig PON, People started to pay for that in commercial developments and it will come 50 gig and 100 gig. And when they come, we will be first as we normally are. On top of that, it is very important to have management platform that support the current needs of the industry in fixed Corteca for the devices and Altiplano for the network is helping our customers to make more money, to monetize their investment, but also to save some money. And -- because in fixed at the end, it is about investing in innovation to be ahead of competition all the time and investing in cost improvement to match this end customer prices demand that I was talking about. In IP Networks, we are investing, doubling down in data center for AI, as you saw in our last quarterly results. We have been slowly but surely positioning ourselves in that domain over the last 3 years. And now we're starting to get some benefits out of the data center space. In the normal routing space, we are very happy to see that in Q4, we were #1 worldwide in total routing, excluding China. And in North America and in EMEA, we're also #1 for the first time ever, for the first time ever globally. Again in EMEA and for the first time ever in North America in total routing, which is -- I mean, this has been a ride of 20 years before we got into that position. But now we are leveraging that position to double down, as I said, in investing in data center and opening new spaces. And in data center, also it's about innovation. We just launched our EDA platform, which is the event-driven automation platform, which is as important as the hardware and software liners or whatever other software we are using itself because in data center, it's not just about getting the right cost point with the right level of performance, resilience and quality, but also to have the automation tools that allow you to manage 100% automatic all the things that you have in the data center. So automation is absolutely key. So innovation is what is giving us the possibility to -- even if we recognize we are a challenger, we are gaining momentum more and more in the data center space. And we have already some recognition by the market with some references in GPU as a service type of networks with CoreWeave and Nscale and hyperscalers like Microsoft Azure, who selected us for their data center deployments. On top of that, we are in partnership with people like Kyndryl or Lenovo for enterprise data center. And of course, in service providers, we have very good references as well when the service providers are building their own data center for their consumption or for to provide AI as a service to others. Now the industry is recognizing us, for the second year, GigaOm recognized us as an innovation leader and outperformer in data center switching. So we are positioning ourselves as I said. And finally, in optical networks, we have a supercharging business now with the combination with Infinera, we are surely excited about this combination because it's going to allow us, well, to have the volume and to have the scale and to complement ourselves in order to better address the market. But I'm not going to spoil the presentation of James Watt. So I'm going to give him the floor, and he will talk to you about the optical network.

Thanks, Federico. Thanks, everyone, for being here. So maybe before we get to optical networks, I back up a bit to what's going on, what AI is bringing to the optical part of the infrastructure, which, of course, as we all know, is where things start and some days even where things end. So it's bringing a lot of pressure, right? There's always something that's transforming the optical networks. AI is bringing a lot of pressure. Three main areas, a lot of others, too, but 3 main areas: need for capacity, not just capacity, but capacity in the right places in those networks, right? There's a push for capacity on moving training data sets around. There's push for capacity on getting the inferencing answers to the right place and getting things between the different elements of that. So capacity, capacity in the right place. Power is another one. We've been hearing about this now for a while. This isn't new. Even the optical networks space, we've been working like crazy to take power out of the network for the last 15 years ever since we started moving to coherent. But the power constraints and the impacts of power in the AI world are not just on the equipment themselves, but they're on architecture, they're on latency requirements. They have network level impact as well as the equipment level impact. They're changing where people locate things that has impact on how we -- how networks get built and what we need to give to our customers to build them. And then on the line system side, the great news is with a lot of hard work over the last 1.5 decades, we figured out how to get to Shannon's limit. We can fill up wavelengths, we can fill up fibers. The challenge is all this capacity has to go somewhere. Now we need to find better ways of building networks that get that capacity they need and make using multiple fibers much more tractable than it has been. We've done a great job of getting from C-band to C+L. Now we've got to go beyond. So a lot of pressure on capacity. It comes out in different ways and different places. And if we talk about places, different parts of the market, right? So the optical systems market, we know, coherent pluggables, a massive growth area in the market and for us. And inside the data center, all that training and inferencing, which I was talking about of having impact on the wide area networks, has massive impact on the need to scale inside the data center. And that power equation has to be solved there as well, something that different technologies, different players can make an impact on. If you look across all the parts even of the WAN, different parts are feeling the pressure in different ways. DCI, a lot of pressure on that, different customer bases, a lot of pressure on wholesale and other related markets, MOFN-type markets, people needing connectivity where they don't operate. Submarine links, if you look at the growth for the last few years, I was -- had the pleasure of working beside a division whose whole focus was on subsea cables, and they were going crazy. If you look at the amount of growth, the amount of bandwidth. And when that bandwidth, not only does that bandwidth have to be terminated, when it comes to shore, it has a huge impact on networks. And finally, you see geographical differences why? Different players, different levels of starting point in serving, much like Federico mentioned on fixed networks, where you have places that are underserved, you get a massive growth rate as you catch up on serving them. So significant growth, a lot of it driven by different pressures and you start to see it. So imagine you've got a multidimensional cube here. Some of them are bright green, some of them are just normal green, but there's a lot of different activities that go on underneath. And what do we do with that? Different activities, how do we bring it back to something we can reason about, we can explain to our customers, we can keep ourselves organized and make progress because progress and the rate of progress is the key thing. And going back to scale, right, the reason for scale one of the main reasons for scale for us is to get more innovation that matters in the hands of our customers sooner. But to do that, not only do we need to be laser-focused on what they need, but we've got to have a systematic way of thinking about what we do. And our solution principles are very straightforward. It's about scale, it's about simplicity and it's about security. Scale is, if you will, a bit of a given if you look at the drivers of the market, although I will hasten to point out that, again, as you look at different parts of the network and getting bandwidth to different parts, it's the right scale. It's not only scale up, which is a hard enough problem, and Ed will talk a little bit about that later, but the right scale to get the bandwidth to where it needs to. As these networks grow, not just in capacity, but in size, automation is essential to make their operations tractable. The one thing I'm certain about any network and probably the only thing I'm ever certain about a network is something somewhere needs attention. And the key thing is to make sure people understand that, that the machines, whether it's with AI, whether it's with other cleverness, can tell the humans where they need to intervene if the machines can't take care of themselves. Automation is the only way to make things tractable. And security, I think we hear enough every day, and it's really what we can do to make networks secure and fit into the security mechanism to the left that makes a huge difference in the overall safety and security of the services our customers deliver. Again, small set of principles keep us, especially as a larger team, laser-focused on what we need to deliver to our customers. And where are we starting? Both teams, both sides of the team because it's one team now, have a great track record of innovation. And whether that's the first-ever 100-gigabit single coherent interface deployed, by the way, I'll point out 15 years ago, whether it's PCS, whether it's multi-vendor operable PCS as we're demonstrating down on the show floor, there's a huge amount of technological innovation. And together, this pool is now available to our customers, the results of this innovation pool. And together, we're going to be able to do more faster because we can focus on taking this innovation, and we only need to apply it to the same problem once, not twice the way we've been successfully doing it for the last couple of decades. So more innovation in our customer hand, faster. And if you want to understand what -- where are the building blocks we're talking about. Today, we have not just the biggest, but the best DSP team in the industry at a time where the pressures on innovations, the DSP space, the pressures for use of coherent across different problem spaces is bigger than ever. If you look, we have fundamental research in Bell Labs. If you think about what some of that is and how we all in the industry benefit from that on basic building blocks, we still continue the research to make sure we have what we need to benefit from in the next years. On the material science, which is a key underpinning element of the optical networks industry, we now have the world's richest set of internal resources, not -- most importantly is the addition of indium phosphide to what was already a rather robust set of capabilities, not just in the material science and the design expertise, but also the fabrication, test and packaging of those devices. So we've taken 2 very successful teams. Together, we have exactly the set of technologies that you need if you go and revisit the needs for scale, for power, for latency and for the architectures that are going to underpin the AI networks. And that's, if you will, where we come from and what we brought together. The other part of this is, again, the people, but the other set of people, the people that take these innovations and bring them to life with our customers, in fact, drag out the need for innovations from our customers, take them, execute projects, take care of our customers and evolve their networks. And that's the other part of human capital, the local expertise delivered globally that complements the innovation and make sure we get the most and our customers get the most out of it, meeting the challenges they have. If you take that, you look at us today, we're -- we are the only player who's #1 or #2 in all markets of the world, right? And that's -- we're only getting started. We have a lot more to do from here, but we have a great starting point, a great ability to take care of our customers and make sure they get what we need. So overall, where does that leave us today? 1,000-plus customers, roughly $3 billion in revenue, a massive amount of traffic carried across the top -- 9 of the top 10 hyperscalers, 48 of the top global service providers and ae massive set of mission-critical enterprise customers. So again, multiple market segments, multiple technology segments, underpinnings of the networks that keep things going day in, day out despite the pressures, which continue to come at them. Now that's where we're starting. As I said, there's a lot more to write. So if you think about -- I talked about innovation, let's group that into 4 buckets. right? Platforms and optical line systems. So really the photonic infrastructure, automation to make it tractable, as I said, the coherent optical engines that get the bandwidth across that, but also getting the bandwidth in intra-data center. So Ed, if you want to pick up on the first couple, please?

Thank you. Good morning, everybody. Great. Thanks. So I get to talk a little bit about systems and automation and security. So first of all, I want to talk a little bit about the portfolio. When we talk about systems, it's really the boxes we sell. And pulling these 2 portfolios together give us a full suite of systems that we sell. First of all, with OTN switching, this is really optimized for efficient large-scale grooming, large-scale aggregation in the large CSPs, mainly in the core of the network. And we have a whole portfolio going up to 48 terabit of size there. In the metro and the regional space, we have packet and OTN. Here, the ability to aggregate and groom any type of data traffic coming in, be it Ethernet, be it some legacy protocols, anything as well as really security of delivery of those services really focused on latency, really focused on size, really focused on density, okay? And then the last one is compact modular. So really efficient, very high transport capacity focused on data center, and that's data center form factors, data center applications, front-to-back airflow, various things like that. And how do we get really all of our electro-optics engines or coherent engines, as James referenced, into all of these different boxes. And we pulled all of that together, and it's really a strong suite of ours. When we talk about OLS and OLS' optical line systems, those have been evolving a lot over the years, right? C+L is very standard today. We introduced L-band, wow, as part of Lucent back in 2000, but C+L we've had out for a long time. As capacity continues to increase, the capabilities in the OLS and the ability to route any wavelength anywhere or come up with the right density for a particular application is really, really critical. And we've got a full suite here, which is really nice. But fiber capacity, the different architectures for the OLS are -- are going to continue to be growing points for everybody, especially as we're at Shannon's limit now. If we look a little deeper, I talked about C+L. That's really the standard that's being deployed right now. And this isn't just in the long haul like it used to be, but it's in metro, it's in regional. It's really everywhere. And we have purpose built around this, some for CSPs that allow us to grow as their traffic patterns grow, so they have a more pay-as-you grow, but also combined C+L, so they can put everything in day 1. Super C and L-band gives a little bit of capacity increase for some particular applications, really nice for some of the wholesale space and things. And then we have hyperscale OLS. All of our hyperscale customers are driving slightly different architectures and are looking for different things. But there, it's really very high capacity, very dense solutions that allows them to deploy a lot of traffic day 1, in essence, and continue to grow that, okay? And then where are we going? There's a lot of work going on in terms of multi-rail OLS. And what do we mean by that? You've heard in years past at this conference about SDM, okay? Multi-rail is really the step to get us there from a node or a network perspective. How do we make sure we have the right components so we can transmit over multiple parallel fibers, okay? And that could be bundled, that could be multi-core, various things like that. But how do we continue to consolidate and integrate our equipment to really optimize those solutions as customers are deploying more and more parallel waves. The other key one is hollow core. And this really brings some key application advantages and some real interesting technology advantages that will really drive the OLS and where we're going with the OLS. So first of all, with hollow core, significant latency improvements, okay? Second is much higher power you can transmit over this, which gives you some better reach. And more importantly, is significantly lower electro-optics or impairments based on the fiber, which again gives you better performance and other things, okay? But there's things we're working on to really optimize the overall line systems to go into all of these. With all that great systems work, all of that really comes together with the software, okay? And that's the automation software. So you see in this circle here really what we call the life cycle of the network. You got to plan it, you got to commission or deploy it, provision. And as James said, you got to assure that it's working. That's the main thing is make sure, one, customers can grow quickly, make sure that they can really monitor and the system identifies and corrects where needed, faults as quick as possible. And then it's optimized. How do we make sure we're packing enough wavelengths onto each fiber? How are we making sure that the customer can grow quickly enough? And that's really from our overall suite of products in the network automation area. Now AI, James talked about how AI is driving capacity. It's also driving some of the capabilities in our products. And we're always looking at ways to automate overall. With the bandwidth continuing increase, it does bring more complexity in terms of how the customers are deploying it. And thus, automation is becoming more critical and more mandatory throughout, okay? We're looking at automation for things like sensing and detecting patterns in the overall data streams so we can predict when there's going to be a fault or when there's going to be a fiber cut, okay? And you guys have seen that at some of our past shows. We've got that ongoing. We're using AI to help with diagnosing problems and not just raising alarm like used to happen, but really diagnosing and projecting or suggesting to the customer what the overall resolution is. So all of that ties together. And really, automation in our overall network automation area is critical. As we're all here in San Francisco walking around, we see the Waymo fully automated cars driving around. Our plan here is really fully automated networks, okay? And really tie that all together. And truthfully, there's nothing preventing us from getting there. It's just working down the path to get there. When we talk about security, similar things. This is really coming or becoming a bigger and bigger buying decision for our customers. So with many of the customers, it's the government certifications or security certifications like ANSI-NIST Common Criteria, but also signal path encryption, and we support this in multiple ways. But with the introduction or the evolution to quantum computing, what's becoming more and more critical is really quantum-safe networking. How do we make sure that what we're transmitting today in those data streams is safe for the future? There's a lot of concerns about people recording data today, storing it to be able to decode it later, okay? So we do a lot of things around symmetrical pre-shared key distribution, rotation on periodic basis as well as secure hardware to make sure even when quantum computers become more and more popular, that is not a key aspect here. And this is really a key thing for many of our enterprise customers and CSP customers, okay? So that's the gist around the systems and automation. I'm going to hand it over here to Ron.

Thank you, Ed. I'm really excited to be here today and be able to represent the subsystems piece of bringing the optical teams together. Four weeks ago, we closed this acquisition. And day 1, when my badge said Nokia, we showed up in Murray Hill, what an amazing place to show up to have this discussion. But we showed up in Murray Hill, where Ed and my team were able to get together, compare all of our notes, all of our customers, all of our plans and bring the 2 portfolios together and make decisions about what we were going to do going forward in order to maximize our ability to move very, very quickly and to really focus on a few key things: time to market, cost, performance and scale. And I think that, that is something that I'm really excited about because bringing these 2 teams together I would say -- I would argue that with the way that the market was going, with the fragmentation that's happening with the content providers, it really made it so that we had to build more DSPs to be relevant in more applications. And bringing these 2 teams together and having the amazing capabilities in both Infinera and Nokia to be able to pull those together really gives us this opportunity to hit the market fast and to do that in a way that allows us to go after all those different applications. And that's what I'm going to tell you about. In this subsystems team, we have the ability basically to build optics. Those optics are really coming in 2 capabilities now, one with silicon photonics, which has some unique capabilities for certain applications and with indium phosphide, which has a different set of unique capabilities for certain applications. We have both of those now in our toolbox. With that, we build optical engines. Those optical engines sit in front of the DSP to modulate the light onto the fiber. And we have the DSPs. This is a critical area. DSPs are complex. They take large teams. They take a lot of money to invest in. So having the revenue to justify the R&D spend to go after the multiple DSPs that are required to hit all the applications, and we'll talk about in a minute, is super critical. And I think that's -- we bring a unique capability now to have the scale and the team to be able to execute to that. We take these optical picks and we put them in pluggables, and we drive those towards the data center. And you could see that in our booth downstairs with our ICED portfolio. We also take it and put it in other pluggables with coherent that allows us to leverage those pluggables directly into systems like what Ed talked about or into -- directly into routers. A lot of efforts gone into integration into all the different routers within the industry to go after the ICP wins that we talked about earlier in the year. And then we also embed those into larger engines where we throw more power at the problem to get better spectral efficiency. And so all of these capabilities exist today. They're all shipping today, and we have the opportunity to now move faster towards the next variant of these different engines in the next couple of years. Those engines, whether they're pluggables or embedded, fit into all of the systems that Ed spoke about. And going forward, we will also be able to take these coherent engines and put them directly to routers. You'll see the Nokia router downstairs that has the ability on a per slot basis to do 36 ports of 800 gigabit, an amazing amount of traffic, amazing amount of capacity in a 1RU footprint. So what's really interesting now is that because these applications are so prevalent, because they're so necessary, all that AI traffic that's happening inside the data center, driving massive needs for scale-out applications, 500 meters, 2 kilometers, but it's also spilling over into the WAN. And as it spills over into the WAN, we really see an opportunity to optimize around what's called campus networks. This can be anywhere from a few kilometers to 20-kilometer type distances, sometimes impacted by the use of optical switching and the capacity or the implication that the insertion loss has of putting optical switch in the path. Scaling up to access, metro, DCI is kind of a generic word. It's used to really talk about any of the connectivity in the content provider, long haul and also in subsea. So when we think about this, it's optimization around in scale-out applications, 500 meters to 2 kilometers, 10 to 20 kilometers, 100 kilometers, 1,000 kilometers and then beyond. So we have a need to have optimization around all of these. That doesn't mean we build that many DSPs. What it means is that the combination of DSPs and optics drive the optimization for each of those areas. The last piece I want to speak to is what we call ICED. ICED is the ability to take our indium phosphide integration and build an optical front end for multiple different types of reaches and multiple capabilities within the data center. The most interesting one is the one here at the bottom because what that really does is if you have a captive environment where you control both ends of the circuit, you can get a massive amount of power savings by eliminating the DSP from the optical pluggable and use the switching chip or the driver from the switching chip to drive that directly. We call this linear drive optics. And then we can also leverage that same engine to go into half retimed or fully retimed applications. So these give you either multiple different reaches or different compatibilities with other solutions. What you'll see in the booth is that we take this engine in this kind of format, and we could drive it today even 15 kilometers. So it's really an interesting evolution in this space where some people talk about coherent light, like taking coherent and bringing it down to 20-kilometer type applications. There's also a push to move IMDD or these type of intra-data center applications up to higher distances. So as we can push this up, it's like electrical versus optical in building a system. If you can make electrical work and you can scale it, the answer will always be electrical. If we can scale this up, this will always be -- IMDD will always be the solution to do that. Ultimately, as we transition from this generation, which is 1.6 terabits and beyond, we go to 400 gigabit per lane, this will drive the need for coherent to get to these type of 20-kilometer distances. But the goal is always to push this as far as we can for these different applications. And with that, I will turn it over to James.

Thanks, Ron. So if we look across all that, where are we ending up? Where are we going to be? Innovation is the core of it. And that is more innovation sooner, more in the hands of our customers. On scale and breadth, yes, it's the scale of the business, which, as Ron pointed out, has an effect of its own, but it's the breadth. As you saw, not just scale in the core optical networking business that we've been in, but scale into the pluggable optics market and even beyond applying our technologies, our capabilities to increasingly other markets and taking our track record of innovation, of problem solving and delivering reliable solutions and using that to help our customers solve the problems they have. And that's the optical networking part. But as Federico said, we live in network infrastructure. And the amazing part of that is I have on one side of me, the world's best IP business. On the other side, the world's best access. And together, we are the preferred network infrastructure provider for everybody, everybody around the globe and growing that infrastructure that we can provide into the data center while taking great care of our core markets. So overall, that's where we are bigger, stronger and faster. Thank you.

All right. So that's the presentation part. By the way, I don't know if you guys have been to the show floor or not. We do have two booths. So we are -- some clever people came up with the line too much innovation to fit in one booth, which is absolutely true. It's a tremendous amount of capability of showcasing all the way from the IC pluggable to a line system running over hundreds of kilometers. So really you should go check it out. We have our IP, our optical and our fixed access solutions, the entire NI portfolio on display. So please come check it out.

We're going to start the Q&A session. We have everybody available here. So whoever wants to start, we have a mic, I guess, or you guys will have to just speak up. But we have mics. All right. Mic coming. Cool. So 1, 2, 3. James, do you want to come to the middle a little bit?

Simon Leopold with Raymond James. Two questions that I think maybe are a little bit related. One of which is, I think one of the challenges that Infinera had faced as a stand-alone company was driving utilization of the fabs higher. So I'd like to get a better sense of what Nokia can bring in to improve that utilization of the indium phosphide fabs. The other aspect is coming out of meetings at Mobile World Congress, which was like right after the deal closed, I felt like ICED was considered sort of a call option wasn't -- was sort of above and beyond the expected synergies. And in this presentation, it looks like it's a real part of the strategy. So did I misunderstand the message at Mobile World Congress? Did it evolve? And any of the above?

I'll take the second.

Okay. Go for it.

Yes. I mean, what can you expect? When you do a business case to propose your Board to make an acquisition, you prepare for the worst and hope for the best. And you don't take into account everything that at the time that I proposed to the Board was not so clear as today. But of course, now we are after day 1, and now it's part of our business. That's what it is. It's an upside that we're going to try to leverage as much as we can, believe me.

And to the former -- the first question, Simon, I think there are 2 parts to that. One is simply scale, right? There -- the increased volumes will increasingly fill the fab. But also as we spread -- as we apply the technology to other problems, it will also drive incremental volumes as well. So I think the -- as I said, there's a scale part, but also as we broaden out the business, that will bring more to it as well. ICED being the first but not only example.

Second one here, and we can go third there.

I actually have 2 [indiscernible] just really 2 questions as well. First question is on Ron, on your presentation about IMDD instead of Coherent light. And then you also had on that slide 8, I think, 8 watts of LPO. I thought the underlying story there was you use Coherent light because like the hyperscalers who want to move things closer on the campus distances of under 20 kilometers. I haven't seen like IMDD being really talked about in that sense. And then within that, LPO is supposed to be getting rid of the DSP. Why do you have a solution with the DSP in it?

I think the interesting thing about the -- I'll take the second one first. The DSP or not the DSP really has to do more with interoperability and the ability to go after the whole market in terms of intra-data center. So we want to have that flexibility to go after the whole market. Some of our customers have aspirations or have the capability today to have an optical switch in the path. Sometimes that requires more reach, more DBs to be able to go through that optical switch and deal with not just one hop, but a hop back to the switching elements or the AI elements. So it's optional, right? And the goal of the technology was to build a front-end optical engine that's flexible enough to work with a DSP or without a DSP so that you could, in some cases, get this great performance without a DSP to enable not just NIC-based LPO solutions, but also switch-based LPO solutions. And then with the DSP, there's been a lot of work by some of our content provider customers to work with the Marvell's or Broadcom's or Credo's or MaxLinear's of the world to drive those capabilities that they want inside the DSP, and they dictate that. They expect that in their solutions. So we don't want to preclude ourselves from being able to apply to those. And I'm not sure you guys in the room that predict LPO versus timed have got it right, right? I think that as time goes on and people get more confidence in LPO, we'll see more of the market TAM move from timed to LPO. But we just don't want to preclude ourselves from both.

And the go-to-market for that is building your own modules or kind of like doing GDM stuff like what Marvell's, Broadcom's and all these guys do.

Yes.

Okay. So it's building your own modules.

No, it's all -- it's whatever path to market makes sense, right? And there's multiple paths to market that makes sense. We're not going to be the lowest cost module provider when it's really, really simplified or when our content provider customer has -- is dictating that it has to be in a light or it has to be TFC or has to be source photonics, right? We're just going to meet them where they want to be met.

Third question here.