Good morning, ladies and gentlemen. Welcome to Nokia's second quarter 2026 results call.

I'm David Mulholland, Head of Nokia Investor Relations, and today with me is Justin Hotard, our President and CEO, along with Marco Wirén, our CFO. Before we get started, a quick disclaimer.

During this call, we will be making forward-looking statements regarding our future business and financial performance, and these statements are predictions that involve risks and uncertainties. Actual results could therefore differ materially from the results we currently expect.

Factors that could cause such differences can be both external as well as internal operating factors. We have identified such risks in the risk factor section of our annual report on Form 20-F, which is available on our Investor Relations website.

Within today's presentation, references to growth rates will be on a constant currency basis, and other financial items will be based on our comparable reporting. Please note that our Q2 report and a presentation that accompanies this call are published on our website.

The report includes both reported and comparable financial results and reconciliation between the two. In terms of the agenda for today, Justin will go through the strategic highlights of the quarter, and then Marco will go through our financial performance.

We'll then move to Q&A. With that, let me hand over to Justin.

Thanks, David, and hello, everyone. Our second quarter showed continued progress against the strategy we set out at our Capital Markets Day.

Our team is focused on maximizing our opportunity in the AI supercycle, and that focus is translating into early results. I'm pleased with the progress the team Nokia has made in the first half of 2026.

In Q2, net sales grew 9%. We expanded our gross margin by 70 basis points to 46%, and our operating margin by 70 basis points to 9%.

Network Infrastructure delivered strong growth, led by Optical and IP Networks, with sales from AI and cloud customers more than doubling year on year. Mobile Infrastructure sales also grew, and the business delivered stable profitability, largely driven by product mix.

Marco will take you through the details of our financial performance in his update in a moment. I want to take a step back and look at how our first half performance demonstrates progress against the strategy we set out in last November.

As a reminder, these are the five priorities we shared at our Capital Markets Day, and I'm pleased by the progress we've already made across each of these areas. Let me touch on a few highlights from Q2.

AI and cloud was the strongest growth driver in the quarter. Net sales more than doubled year on year to €446 million, and order intake grew to €2.8 billion.

While we're very pleased with the order growth, it's important to put that number into a bit of context. Q2 benefited from several significant long-term orders, as our customers looked to secure supply in a constrained environment.

To provide some reference, approximately half the order volume received in Q2 is expected to convert to revenue in the next 12 months. As I've said before, order patterns in this market can be lumpy, and we should not expect this level of intake every quarter.

As importantly, the strength was broad-based across Optical Networks and IP Networks, and included some of the design wins we mentioned last quarter. This was driven by growing demand for data center interconnect and scale-across fabrics from our customer base.

The demand primarily shows up in our AI and cloud segment, but we're also seeing emerging signs of growth in telecom customers as they invest to support the increased data traffic driven by the AI supercycle. During this quarter, we also secured our first multi-rail ILA design win with a major customer.

This was one of the new optical networking products we launched at OFC this past March. Last week, we launched the industry's first commercial AI-RAN platform, marking a fundamental shift from a hardware-defined radio network to software-defined platforms.

This fundamentally changes the economics of radio networks. Our AI-RAN platform gives our telco customers a path to improve network performance through software and AI innovation, rather than relying on hardware upgrades as they have traditionally.

The platform would deliver more than 100% spectral efficiency gains by 2028, doubling the capacity operators can get from their existing spectrum. The performance benefits are tangible in 5G networks, and our AI-RAN platform provides a software upgrade path to 6G to ensure continuity without additional hardware investment.

The platform is also open, programmable, and O-RAN compliant. This gives operators greater flexibility as they evolve their networks.

They can choose the hardware path that works best for them: adding AI acceleration into their existing Nokia AirScale infrastructure, deploying new AI-RAN hardware, or moving to cloud-native AI-RAN. Ultimately, this is about delivering more performance, better returns, and faster delivery of new service for our customers.

We're on track to enter pilot deployments at the end of this year, and expect to be commercially available in 2027, as we've said previously. Co-innovation is a powerful differentiator for Nokia.

When we combine our technology leadership with the expertise and scale of our customers and partners, we accelerate innovation, bring solutions to market faster, and solve increasingly complex challenges together. We're already demonstrating early results from this approach, and I will highlight four examples from Q2.

First, we expanded our partnership with Google Cloud, bringing Gemini-powered AI agents into our autonomous networks portfolio. Second, with Vodafone Albania, we demonstrated AI-powered network slicing, using agents to dynamically optimize network resources.

Third, we expanded our relationship with Indosat Ooredoo Hutchison in Indonesia, supporting network modernization and the rollout of 5G, while providing a seamless upgrade path to AI-RAN. Fourth, we entered trials with a US hyperscaler for a new out-of-band management solution that goes inside the data center, leveraging the passive optical technology that we deliver in our Fixed Networks business.

We're also making progress to focus Nokia where we can differentiate and create long-term value. This means we are investing where we see long-term demand and we believe Nokia can be a unique winner, and at the same time reducing exposure to areas where we are less differentiated.

In November, we shared that our Fixed Wireless Access customer premise equipment portfolio is not core to the future of our strategy. The sale we announced this past quarter to Inseego is an example of our disciplined approach to capital allocation, and allows us to concentrate resources on higher-priority opportunities.

The sale is on track to close by the end of the year. Talking about higher-priority opportunities, one area of focused investment is scaling the capacity needed to support our optical growth ambitions, particularly around indium phosphide semiconductor manufacturing.

In San Jose, our new indium phosphide fab is now processing test wafers, as we move closer to product qualification. It remains on track for volume production by the end of the year.

In June, we announced a new commitment we are making to scale our Pennsylvania facility, increasing our advanced test and packaging capacity for optical systems in that facility by 10 times. In addition, today we announced the acquisition of a manufacturing site from NXP in Arizona, where we plan to increase our indium phosphide fab capacity.

This gives us additional capacity to support our own demand and greater optionality, recognizing the supply constraints in the market today. Together, these investments continue to strengthen and secure US-based optical manufacturing capacity for the long term.

While Marco will update you on our restructuring progress, I wanted to touch on one key area where we are making progress in driving incremental productivity. We believe that to be a relevant technology provider in the AI supercycle, we need to be a leading adopter of AI internally.

Last year, we established a team to deploy AI test beds across multiple functions within Nokia. One area where we're seeing early traction is software development, where we now have nearly 100% adoption across our developer base.

This is already yielding significant productivity returns, supporting our efficiency targets, and accelerating road map deliverables. We will continue to scale this initiative across every function of the organization as our test beds yield tangible results.

We see this as essential not only to unlock sustainable returns, but also to be a better partner in support of our customers as we help them unlock the full potential of the AI supercycle. In closing, I want to recognize and thank team Nokia for a strong first half of the year.

We are focused on our key priorities and have begun to fundamentally change how we work. I'm pleased to see our efforts are already reflected in our results.

We're entering the second half with good momentum and remain on track to deliver somewhat above the midpoint of our operating profit guidance. Now, I will turn the call over to Marco to dive into our financial performance.

Thank you, Justin, and hello from my side as well. Before looking at the quarterly performance, let me start with reporting changes we announced this morning.

As we have agreed to sell our Fixed Wireless Access business to Inseego, we now consider the sale of Enterprise Campus Edge highly probable. As a result, both businesses are classified as discontinued operations.

We have published recast historical numbers for 2025 and quarter one 2026 to support comparability. In quarter two 2026, this reporting change reduced comparable net sales by €66 million and increased the comparable operating profit by €13 million.

It also led to minor adjustments in cost allocations between Network Infrastructure and Mobile Infrastructure, with an impact of approximately €1 to €2 million per quarter. Turning to performance, net sales grew 9% in the quarter, supported mainly by Network Infrastructure.

Gross profit was €2.2 billion, and gross margin increased 70 basis points to 46%. The margin expansion was driven by Network Infrastructure, and particularly Optical Networks, where we continue to benefit from both strong demand and the integration of Infinera.

Operating profit was €434 million, and operating margin increased 70 basis points to 9%. The quarter benefited from some software revenue recognition coming in in quarter two instead of quarter three.

We also incurred higher stock-based compensation expense, which represented 150 basis points headwind to our operating margin in quarter two year on year. This was driven by Nokia's share price increase, an increase in the program, and the issuances happening earlier in this year.

Financial income and expenses benefited from a positive venture fund revaluation during the quarter, which supported both net profit and EPS, earnings per share. Free cash flow was negative €732 million.

As quarter two is typically the weakest quarter for cash generation, as employee cash incentives are paid in quarter two. We also saw some increase in working capital during the quarter.

We ended quarter two with a net cash position of €2.8 billion, maintaining a strong balance sheet and significant financial flexibility. Let me now turn to Network Infrastructure.

Net sales grew 12% in the quarter, reflecting continued strength across the business. Optical Networks grew 20%, and growth was supported by continued demand from AI and cloud customers, but we also saw healthy demand from telecom customers investing in transport infrastructure.

IP Networks grew 16%. The strong order momentum that began in the second half of 2025 is now translating into revenue growth.

Fixed Networks declined 2%. The areas where we are prioritizing investment performed well.

Optical line terminal sales grew 18%, while ONT sales declined 16% as we continue to focus on higher-value parts of the portfolio. Gross margin increased 240 basis points to 42.7%.

This improvement was driven by three factors. First, we benefited from higher scale as revenue increased.

Second, we continued to realize synergies from the Infinera acquisition. Third, we saw more favorable mix within Fixed Networks.

The gross margin improvement was partially offset by growth investments we are making across Optical Networks and IP Networks as we position ourselves to capture the long-term opportunity in AI infrastructure. Finally, operating margin increased 170 basis points to 8.1%.

Turning to Mobile Infrastructure, net sales grew 7% in the quarter. Core Software grew 1%, Radio Networks 7%, and Technology Standards increased 15%.

Technology Standards benefited from signing a few new agreements during the quarter and included some catch-up revenue recognition. Looking at the full year, we continue to expect Technology Standards to deliver a similar level of sales and profitability as in 2025.

Gross margin was 49.3%, which was somewhat better than we expected entering the quarter. The main driver was a higher contribution from software sales, as some revenue were expected in quarter three ended up benefiting quarter two gross margin.

Looking ahead, because of the earlier software revenue phasing, we currently expect Mobile Infrastructure gross margin in quarter three to be closer to 44% to 46%, reflecting a lower software contribution before improving again in quarter four in line with normal seasonality. Operating profit was stable year on year.

Looking at sales by customer segment, AI and cloud was again the fastest growing segment, with net sales increasing 105% year on year. Growth was broad-based across both Optical and IP Networks.

Telecom sales increased 4%, while Technology Licensing grew 15%. We remain optimistic about the long-term AI and cloud opportunity, and continue to see strong customer demand.

At the same time, our expectations for the telecom market remain largely unchanged. Turning to restructuring and integration costs.

First, we are on track to complete our 2023 to 2026 restructuring program this year and achieve €1.2 billion in gross cost savings. The second area is the integration of our Chinese operations into Nokia's global operating model after taking full ownership at the end of 2025.

As we discussed previously, we continue to adjust our operating structure to reflect market conditions and improve competitiveness. As a reminder, this program was expected to achieve €200 million in cost synergies, with one-time charges of between €350 and €400 million over a two to three year period.

We now expect to recognize approximately €350 million of the planned one-time charges by the end of 2026, as we accelerate the integration to complete it within two years. The third area is a set of new efficiency programs, mainly impacting Europe.

These programs are expected to lead to restructuring charges of €200 million in 2026. These actions are focused on simplifying the organization, improving productivity, and ensuring resources are aligned with our strategic priorities.

Overall, we expect restructuring charges of approximately €800 million in 2026. Then let's go to cash.

With respect to cash flow, the quarter followed normal seasonality we typically see in quarter two. The largest impact was the payment of annual employee incentives related to 2025 performance.

We also saw some working capital buildup during the quarter, reflecting the continued growth of the business. Despite these seasonal factors, our overall cash generation profile remains unchanged.

Considering some of the increased restructuring costs, and as we make some investments in working capital to prepare for growth, we now expect to track towards the low end of our free cash flow conversion assumption of 55% to 75%. Finally, turning to our outlook, there is no operational change to our comparable operating profit guidance.

The only adjustment is the technical change resulting from the move of Fixed Wireless Access and Enterprise Campus Edge into discontinued operations. We continue to track somewhat above the midpoint of our operating profit range.

Looking at quarter three specifically, we currently assume a sequential increase in net sales of between 3% and 7%. For operating profit, we currently expect a result broadly similar to quarter two, due to the phasing of software sales in Mobile Infrastructure between quarter two and quarter three, followed by meaningful improvement in quarter four.

This is a combination of the normal seasonality we see in our telco business and the contribution from year-on-year growth in AI and cloud sales. Aside from the discontinued operations adjustment, our outlook assumptions remain largely unchanged.

The demand environment remains supportive, and we allocate capital where we see the strongest opportunities for long-term growth, while maintaining discipline on profitability and cash generation.

Thank you, Justin and Marco. As usual for the Q&A session, as a courtesy to others in the queue, could you please limit yourself to one question and a brief follow-up?

Operator, could you please give the instructions?

Thanks, Dani. We'll take our first question today from Terence Tsui from Morgan Stanley.

Terence, please go ahead.

Yep, thank you very much. I hope you can hear me okay.

I had a question around capacity, and particularly around the four new DSPs planned by the end of 2027. This is actually a big ramp-up compared to the previous run rate.

Can you give us some milestones to look out for and reassurance that this could be achieved? Thank you.

Sure, Terence. I think first of all, we talked about these going into customer trial in '27 and then become commercially available towards the end of '27.

I think the thing I would just emphasize and remind you of on this is that Nokia and Infinera previously were each building two DSPs individually, collectively a total of four DSPs. One of the decisions we made, and I've talked about this a bit as we saw the growth opportunity emerging in Optical, was to maintain the DSP team as is versus reducing them.

The reason we did that was we felt quite strongly, in spending time with our customers, that we could actually deliver more differentiated products to them with four unique DSPs versus the traditional two that we had been delivering in each company independently. That laid out and supported the road map that we launched at OFC in March, and I touched on that a bit in last quarter's call.

The view here is that we think this gives us better market coverage aligned to where the market is evolving, specifically given the amount of investment we're seeing across the different layers of optical fabric, from the scale-across fabric to what we see in data center interconnect to metro network to long-haul transport. Across all of that, we think we have a much better portfolio in 2027 to address each of the unique evolving solutions than we would have had if we had only kept two versions.

Did you have a follow-up, Terence?

A real quick follow-up on the comments around the order book potentially being lumpy. Do you still expect an upward trajectory over the longer term from these levels, albeit with some bumps along the way?

Look, I think two things about this quarter, this quarter's order book. Obviously, if you do the book-to-bill on this, it's a significant jump up.

I think, I think for me, that's a that's a data point around, a little bit of lumpiness. The other is the elongation of the orders, which we've been talk we've been talking about expecting, and we're seeing here.

I think we have to look at it in both those dimensions, and the way I think about orders is I think about orders in a period of time, right? Because it's easy to it's easy to give you a headline number and then say, "Well, the next question is, is that over a a quarter period, a four-quarter period, an eight-quarter, 12-quarter, etc.?"

For me, that's really where we're focused is not, is not necessarily on, are we getting, big order big order pops consistently, but rather, are we seeing the order momentum grow as we as we look at it over a time period? Right now, what we're seeing is continued growth and continued demand in the market.

As I said in my comments, it's still largely driven by AI and cloud, particularly around scale across and data center interconnect. Then we're starting to see some, some emerging growth.

We saw some of it in sales this year in in in our telco customer base, and we believe that's also tied to to AI demand.

Thanks, Terence. We'll take our next question from Simon Leopold from Raymond James.

Simon, please go ahead. Simon, have you unmuted yourself?

There we go. Can you hear me now, David?

Yes, go ahead, Simon.

Great. I didn't expect that button to pop up, sorry.

I wanted to see if you could rank-order and characterize supply chain risks, and I'm thinking about issues like memory, printed circuit boards, and even indium phosphide wafers, maybe a particular focus on that latter one, the wafers, given the factory expansion, whether or not you can get the materials, and then I've got a quick follow-up after. Thank you.

Sure. Look, I think the most if I think about the supply chain risks or the the the time that we're spending in this, first of all, as you rightly point out, it's broad-based, right?

I think we talk a lot about memory, and memory is significant just given the the amount of demand that is in the market. I think this has been talked about quite a bit across, multiple companies and and multiple parts of this ecosystem.

There's clearly, there's clearly constraints there, and then obviously the the the pricing, the significant change in pricing driven by that shortage, which again has been talked about very broadly across the the tech ecosystem. That's probably the one that we see as most significant.

Now, we talked about this last quarter. I think, I don't need to repeat what I said last quarter, but maybe briefly, our focus is on securing supply, simplifying our designs, looking at where we can, we can reduce scope wherever possible on our designs, and of course then passing that, passing that on to customers.

I think if you look at what we said last quarter, there were some companies with us last quarter. It seems like more companies have joined us in in some of the more recent earnings calls now making that that same, yeah, that same comments.

The key thing for me here is is also really talking to our customers, not the AI and cloud customers, they understand this well, but really making sure our telco, our mission-critical customers understand that we have elongated lead times, which means better visibility, better planning, and something that we lead the team with them on. Particularly important in that regard.

Then as you as you touch on, there's a broad base across the board. On the indium phosphide question you asked, Simon, the comment I'll make is, this this new fab is really looking at, is is looking at coming online probably earliest in '29.

If you think about our our capacity, we've got, a significant jump-up with San Jose coming, call it '27 as it ramps volume, right? Manufacturing later this year, volume manufacturing later this year, '27 it ramps.

Then we we kind of line up for a '29 ramp and incremental capacity. As that's kind of the the timelines that you you have to take with these, with with these investments.

I think as as we're looking at it, we're looking out at, at different solutions to get indium phosphide capacity at that time. That's obviously an industry issue, though.

It's something that all of us in the industry need to enable, and it's it's something that, I think collectively we need to solve, even even across, even across the ecosystem.

Thanks, Simon. Did you have a follow-up?

I wanted to see if maybe you could offer us your view on the scale-across market and Nokia's ambitions for this particular application, considering optical and IP routing. Thank you.

Yep. Absolutely.

I think first of all, Simon, I'm going to be a little technical, but scale-across technically was talking about data centers within a given campus area that were strung together as a AI factory. One of the things I talked about on the last call was the demand we're seeing in data center interconnect.

My point in saying that is, some of what we're seeing is just increased data center interconnect. Now, you can call it scale-across or some folks may want to label it one way or another.

To me, they're very different, they're complementary and very important applications, because one is back-end connectivity, which is providing connectivity to expand the back end for scale-out, which is the scale-across fabric; the other is data center interconnect providing higher bandwidth between data centers over a long haul on the front end. Both of those have a routing element.

Obviously, the data center interconnect has a very significant demand growth in routing. If you look at our opportunity and why we're talking about growth in both IP and optical, it's because we're seeing growth in both of those elements.

They are complementary, they are reinforcing. Now, the other thing I'll say is, it's not limited to that.

We are seeing some traction in some of the back-end switching. It's on a limited basis, obviously, without getting into all the market dynamics there.

All of this is encouraging in terms of our focus in this area and the traction we're starting to make.

Thanks, Simon. We'll take our next question from Sami Sarkamies from Danske Bank.

Sami, please go ahead.

Hi. My question would be on your supply capability in Optical Networks.

Are you fully constrained or have you been able to build any inventory during the first half of the year?

I would say, Sami, it's a good question. There are always pockets.

Think of legacy products and those areas where we have some supply, but in general, I would think of us as being constrained, right? When we talk about lead times elongating, it's because we're seeing constraints, and particularly on the leading-edge products.

By the way, I don't think we're unique in that. If you look at our ecosystem, again, I think you see the constraints, and you see that across the component suppliers, some of our peers, et cetera.

Obviously, we're working aggressively on that and maximizing the supply. As I've said as well, if you look at our forecast, what we've included in our forecast is the demand that we have line of sight to shipping.

We recognize even that has some risk because that assumes continuity of supply, no disruptions, everything goes perfectly. When we're thinking about this, we're thinking about it from a constraint perspective, aligned to what we have line of sight to in supply.

Absolutely, if there was more supply, I think we'd probably generate more revenue.

Did you have a follow-up, Sami?

Yeah. Regarding Radio Networks, just curious, do you think you are currently gaining share?

You had 5% organic growth in the first half of the year. I think that's a bit more than your main European rival is having, or is this just timing?

My view on this, Sami, is it's timing. Actually Marco talked about the timing around the software revenue recognition we had in Q2, which is tied to our radio software platforms.

I would call this timing. I also think looking at market share on a quarterly basis in this industry is super challenging to get any kind of good signal.

I think you have to go look at it certainly on an annual basis.

Thanks, Sami.

We'll

take our next question from Alex Duval from Goldman Sachs. Alex, please go ahead.

Yes, thank you very much. You talked about further progress in AI-RAN.

I wanted if you could talk a bit about the timeline for this benefiting Nokia in terms of revenue and competitive position, and what your discussions with telco are suggesting in that area. Secondly, back to the AI side, I wanted if you could give an update on switching and the progress you make there.

Could you help us understand the latest thoughts on switching design wins and when we should expect orders and revenue momentum, given the progress you're delivering?

Let me answer the second one first. In terms of the switching design wins, we talked about this a little bit last quarter.

We said we expected orders this year. We saw a lot of those orders come in in Q2.

Obviously, as you know in the design win process, you start small, you get traction, and then you build on top of that as you validate and execute for performance. We're continuing to drive that across a number of customers.

Obviously, we're pleased with the progress we had in Q2. Then, can you just repeat your first question?

Absolutely, Justin. It was just, you had mentioned further progress on AI-RAN.

I'm just curious how you think about the customer feedback and the timelines for that impacting your revenues.

Look, I think basically, Alex, everything is consistent with what we've said: pilots at the end of '26, commercially available in '27. Obviously, we'd anticipate more significant volume going into '28.

That continues to be our expectation in terms of AI-RAN.

Thanks, Alex. We'll take our next question from Ulrich Rathe from Bernstein.

Ulrich, please go ahead. I think we've lost Ulrich.

Are you there, Ulrich?

Sorry. It takes some time for the button to appear here.

Apologies. I wanted to come back to the very strong AI cloud order intake.

You put it into perspective already with regards to the elongation of the order book, as you call it. I was wondering, in supply-constrained markets, we often do see double ordering, which does create a false signal for suppliers such as Nokia.

How do you see this risk? Are there any specific reasons why this would be an unlikely factor for Nokia?

Look, I think first of all, if you think about the customers, the level of sophistication in the customers that are placing these orders, they get to step back and ask what the incentive is for double ordering. I've absolutely seen this, much like you, in supply-constrained markets that I've worked in in the past, and it's particularly prevalent in markets where you're focused on enterprise customers or you've got channels, because customers tend to diversify and look for allocation.

In this environment, the thing that I would flag is for one of these customers to come in and say, "I'm going to double order with you," when ultimately that goes back to supply of leading-edge silicon manufacturing capacity on optical components that they can actively inspect and we transparently share the progress, the question for them would be, what does it do in terms of incentives? The other thing I would say is we're obviously, as we're making commitments on a longer-term basis, we're expecting those commitments from customers as well.

That's helpful. If I may follow up with one clarification.

What would you call a normal length of an order book? Is it essentially 100% of the orders within the next 12 months, because you highlighted sort of the difference with half of the revenues.

Is 12 months for 100% the quote-unquote "normal" here or not?

Ulrich, that's a good question. I think typically we have seen orders within 12 months in our customer base.

Now, again, there's two factors to this, obviously. One is the growing demand.

Is the fact that AI and cloud is a new segment for us, right? I would say we've had less exposure to this, obviously substantially less exposure to this in the past.

Then the second thing is obviously the supply constraints. I think both of those are factors.

If you think about our traditional business, heavily concentrated with telco customers and then obviously some in mission critical, those orders, we may get a contract award, but we would see orders typically within 12 months, and that's really the shift. That's why when I talk about our telco customer base, one of the conversations we're having with those customers is getting more visibility going forward, because they're also used to giving us fairly short-term, within 12 months visibility, and we need to be planning even further.

This is something that we're with all of our sales teams having this conversation to make sure that we're getting better visibility, not because it implies a commitment, but because the risks given the supply constraints in the industry are, we don't want to miss any of their deliveries, while we continue to support them, given their importance to us as well as customers.

Thanks, Ulrich. We'll take our next question from Jakob Bluestone from BNP Paribas.

Jakob, please go ahead.

Great, thanks for taking the question. You're obviously ramping up in terms of AI and cloud revenues.

I guess we're not yet seeing it in terms of margins. I guess my question is just when, and I appreciate that's obviously as these business scale, margins will go up, and particularly on the IP side, you're just sort of starting to scale now.

It'd just be interesting, how long do you think it actually takes before these revenues become materially accretive?

I think two things, Jakob. One is gross margin, and then the other is operating leverage, right?

As we talked about in Capital Markets Day, we're doing a lot of work at the front end of the three-year period to really set the company up to become more efficient, more nimble, more scalable, and get the operating leverage as we drive growth in the business. That's a key focus, and we talked about that by nature that would be a little bit back-end loaded.

Now, like I said, I'm very pleased with the progress we're making and obviously with the demand accelerating higher than what we said at Capital Markets Day, we've got optimism on progress there that we'll continue to improve. On the other side, on the gross margin side, this is an area where I think we're dealing with just a lot of complexity in the mix, and this is a little bit of supply chain, this is also us, we talked about the focus we're making in FN on exiting low-margin business.

Some of those things are just playing through in the business, and you're not quite seeing it drop to the bottom line yet. Obviously we're very clear on what we're anticipating and based on the assumptions we shared at Capital Markets Day and the progress that we're making ahead of those in terms of revenue growth.

Very clear. If I can just ask a quick follow-up.

Just, you mentioned you've got several customers coming in on the IP side, I think you said. I'd just be interested in understanding just sort of the level of concentration of that customer mix.

Would you say you're well-represented across the different hyperscaler customers, or would you say it's still a relatively narrow segment?

I think we've talked about this a little bit before. It's fairly concentrated today.

That's the way you build the business, right? I think we've got very good partnerships and relationships across many of the AI and cloud players, the hyperscalers.

The focus right now is obviously on making sure where we do have demand that we're delivering it, and we're continuing to innovate for those customers. Then over time, obviously, expanding that footprint.

Thanks, Jacob.

We'll take our next question from Oliver Wong from Bank of America. Oliver, please go ahead.

Hey, guys, thanks for taking my question. My first question is in terms of the €2.8 billion AI orders in the quarter, understand that a significant portion pertains to some of your significant design wins in data center switches from last quarter.

I think what would be helpful is if you could maybe try to quantify or guide us a little bit on kind of how much of the total AI orders this quarter that kind of comprise, just so that we can have a better sense of underlying optical-related demand in the quarter. Thanks.

I think I would say it was driven by Optical and IP, weighted towards Optical, and that's probably not a surprise given the momentum we're seeing right now in that market.

Got it. A quick follow-up.

Within Optical, you discussed briefly about sort of scale-across versus regular DCI. I was just wondering what the composition of demand is right now between the two?

I don't think we're breaking that out right now. I just would highlight that I think there's a significant amount of demand in DCI as well as scale-across, and I think the two kind of get either pushed together or maybe the DCI piece gets underappreciated.

That's certainly where for us we're seeing traction on both.

Thanks, Oliver. We'll take our next question from Richard Kramer from Arete.

Richard, please go ahead.

Thanks. Hey, Justin.

My first question for you is on the AI-RAN transition, your customer installed base runs on Nokia proprietary silicon. Do you see the industry long-term shifting away from that proprietary silicon-based set of solutions, and what are the implications for what is a €3 billion run rate of mobile R&D and Mobile Networks margins for that transition?

Thanks.

Yeah. Hey, Richard.

A couple things on this. I've talked about this pretty openly.

I think we're at a point where the industry has to transition. I think we look at what we see on AI-RAN and the spectral efficiency.

By the way, we'll have spectral efficiency on our existing hardware, we'll have some improvements in software, but they'll be an order of magnitude below what we're talking about on AI-RAN. The second thing here is that when you look at leading-edge silicon, you do the math on the cost of leading-edge silicon, and then by the way the supply constraints on leading-edge silicon, in my mind, this is a very clear industry shift that has to happen on the baseband, and that is a shift to general purpose silicon.

Of course, we're partnered with Nvidia in launching the AI-RAN solution. There are other players out there with general purpose base solutions that are delivering virtualized RAN stacks.

I don't think we're alone in this move. Fundamentally, as we think about the R&D in this space, this is the other thing I've touched on.

When you look at this industry, there's two fundamental challenges certainly from 4G and 5G. One of them is that the cost of capital, the return on invested capital at an operator level in aggregate hasn't delivered, right, in terms of the investment in certainly looking at 5G, but also 4G.

If you look at it from a supplier perspective, a technology provider like us, it also hasn't been acceptable on our side. I think we also have to look at how we shift investment and generate a better return on invested capital.

Getting out of purpose-built silicon on the baseband is a step in that direction, and that's why we've said that's the long-term direction. I think it's incredibly compelling when you can also say to a customer, "By the way, look at the better efficiency you can get on your hardware, which means you're going to get a better return on that hardware investment."

Then the last thing we're saying is, by the way, when we deliver the hardware, that's not the best performance you're going to get. We'll continue to provide performance enhancements.

As an industry, we've always added features, but the fact that we're now adding performance capabilities in our software stack, we think is a huge advantage. The final point I'll make is, as we've talked about in the software stack, this is a single software stack.

We've got capabilities to optimize it for different hardware, including our legacy stack and of course the Nvidia GPUs that are now coming into our portfolio on the AI-RAN platform. It's a single software stack, we're getting a tremendous amount of leverage out of that stack.

This is right on that path.

Thanks. Then a quick follow-up for Marco, if I may.

Your comments about being above the midpoint of your full-year profit guidance, but having flattish profits in third quarter, suggests you're going to more than double profits in Q4. Can you talk through the drivers of that, be it software revenue recognition, licensing, product deliveries, cost reductions, especially given the cash outflows for restructuring, capex, working capital, etc., that we're seeing now and can expect in second half?

Thanks.

Thank you. I would say that if you look the normal seasonality that the industry has, specifically on the telecom customer base side, is usually very Q4 delivery-heavy, and that's why we've seen in the past years as well that that part of the industry usually generates the biggest profits and sales as well in quarter four.

Now in addition to that, we see also increase in AI and cloud customer base that is also impacting the seasonality of our operations and also profit generation. What comes to restructuring the program that we announced in '23, end of '23, that we will end in end of this year, and we expect that we'll get those €1.2 billion accumulated gross cost savings, just like we said as well.

In addition to that, we expect to actually accelerate the synergy program that I mentioned earlier what comes to the China company that we took over 100%. Then also we had some additional voluntary-based cost saving restructuring also in Europe, and those we will take this year as well.

All together, if you look our cash position, we have very strong position. Now we had €2.8 billion net cash end of quarter two.

We had some inventory and working capital increase in quarter two to secure also deliveries going forward. Then of course, accounts receivables follow normal sales pattern as well.

We believe also that by the end of the year we have very good financial position and cash position, I don't see any issues there.

Thanks, Richard. We'll take our next question from Sandeep Deshpande from JPMorgan.

Sandeep, go ahead.

Hi, thanks for letting me on. I want to understand from your order intake in AI and cloud in the quarter.

Sandeep, we're really struggling to hear you, Sandeep.

Can you hear me? Can you hear me better now?

Just a little better.

Yeah. Very strong intake in AI and cloud in the quarter.

How much, last quarter you had said, €1 billion of orders were an ongoing order intake, even though you will have lumpiness in your orders. How much of this €2.8 billion is an ongoing order intake, would you characterize?

Then secondly, regarding AI and cloud, how should we be looking at a run rate on revenue in this business between Optical and IP routing?

Sure. On the first one, I think we've broken it out for you that what we saw on next 12 months and forward, and I'm not going to break it out any further in terms of or try to estimate that for you.

That should give you a good view on what's in the coming four quarters and what's beyond that. Then in terms of the mix, I think I touched on this.

Optical is growing a little bit faster this quarter than IP, but of course it's starting from just a stronger position. IP, we're just, as we said, we're just starting to ramp in design wins and deliver those.

We talked about that last quarter. I'm pleased with the momentum, and I think, if you look at it from the other side, which is 100% year-over-year growth, I think we're set up for a very, very good continued growth forecast from the AI and cloud segment.

Thanks, Sandeep. Did you have a follow-up?

No, I'm fine. Thank you.

Thanks, Sandeep. We'll take our next question from Sébastien Sztabowicz from Kepler Cheuvreux.

Sébastien, please go ahead.

Hello, everyone. Thanks for taking my question.

On AI-RAN, coming back, have you seen a specific commercial traction over the past few months? Have you been added in new customers trialing your solution?

You are targeting twice more spectral efficiency by 2028. What about the total cost of ownership of this solution?

Next, to the baseband, do you plan to partner with Nvidia on GPU for radios, or it will be mostly focused on the baseband? Thank you.

Okay. Three questions.

Let me hit them. I'll start actually with the last one.

The announcement that we made last week was around AI-RAN for baseband and the Nvidia GPU solution going into our AirScale platform, a future standalone platform, and also having a cloud RAN common off-the-shelf server solution. That's the current announcement.

In terms of the spectral efficiency and TCO, as you probably know, that TCO ends up being a very customer-specific discussion. At the macro level, a hardware deployment with a 100% spectral efficiency improvement, I think the math there speaks for itself in terms of the value creation potential for the operator.

The other key thing is that this is a software model, the benefit for the operator is not just TCO, but it's also a capex to opex transition in terms of ongoing benefit without having to have hardware upgrades. I think there's a tremendous amount of value when you look at it from a lifecycle standpoint.

Then in terms of the pilot deployments, we've announced 10 public customers on track for later this year. There's many conversations going on about this.

We expect to start the pilots later this year, obviously expect that will continue into '27. Obviously as we make progress and we should continue to share the progress publicly as we can on the progress we're making, the capabilities we're delivering.

It's more than just spectral efficiency, it's also a platform that's extensible, and we talked about this a little bit. It's a bit technical, but you can actually put your own applications and services in at the radio layer, and this allows some new capabilities which we think are going to be pretty attractive to a number of operators: things like sensing and other applications.

Thanks, Sébastien. We'll take our next question from Rob Sanders from Deutsche Bank.

Rob, please go ahead.

Hi, thanks for taking my question. First question would just be about the indium phosphide fab ramp.

Do you have line of sight to hitting best-in-class 6-inch yields next year? Clearly, Coherent is already doing pretty well; Lumentum seems a bit behind.

Where do you stand on that? The second question would just be around AI-RAN.

If you look at the top three US operators, how many do you think internally have already gone past the go/no-go decision on whether to deploy AI-RAN? Thank you.

Got it, Rob. Thanks.

I I think, obviously, we have one operator today in the US. I'm going to start with the AI-RAN, I'll come back to the indium phosphide.

On AI-RAN, we have one operator in the US that's got our our our RAN deployed at scale, that's T-Mobile. They announced that they're, they're going to be our lead our lead partner on on the pilots.

Obviously, we're working closely with them. I would assume that that would lead us to conclude that they're probably not past the go/no- go/no-go deployment path.

On the others, I I think it's a it's a discussion, that, that obviously, we'll leave for them to assess. My my view here is that the GPU performance is compelling, and it's particularly compelling in a business case where, spectral efficiency matters, which is going to be more dense operations.

That that's obviously, they've got road maps and strategies. Probably better to ask them than than ask me.

Then on the on the indium phosphide ramp, what I would say is we've got yield targets that we've we focused on, both yield and and volume targets we've focused on through 2027 on the fab. My my view is while we're, and you you rightly said while the ecosystem is maturing and it's not just the two you mentioned, but also the Chinese manufacturers in this space, I I also believe this is a place where we're going to go through a significant amount of maturity and learning as we scale and scale yield, and and that's what I'm focused on with the team versus a a specific target or or competitor reach.

I I think there's actually I think this is more about us learning and scaling and and making sure we can we can deliver on our our volume plans and obviously our cost points.

Thanks, Rob. We'll take our next question from Artem Beletski from SEB.

Artem, please go ahead.

Yes, good afternoon, and thank you for taking my question. Relating to AI, could you maybe comment what type of order intake development you actually see outside of AI and cloud, namely telcos and mission critical?

How we should think about the revenue growth trajectory on this front looking at this year and also next year?

Thank you, Artem. Just like we mentioned earlier as well, that we had a good order intake development and sales development in also non- AI and cloud customer base, telcos were investing more.

This is also driving because of their need to invest in their networks to secure that they can deliver the demand that is coming from AI and cloud, and AI in general, development. Most likely, this will happen more broadly going forward as well, because we believe that AI demand will continue, the underlying demand will continue for longer period of time, and without very good secure networks, it is very difficult to provide those improvements that AI is actually providing.

I don't know if you have something you want to add?

No, I would say the only thing I would say, Artem, is if you looked at NI specifically, the only headwind, which we talked about last quarter, is obviously on the customer premise equipment side of Fixed Networks, where we're getting much more disciplined on margin. That creates a bit of a headwind when you look at NI as a whole.

Underneath that is the momentum that Marco talked about in IP and Optical, and also, obviously a healthy growth in optical line terminals as well, which is the network side of the Fixed Networks business.

Thanks, Artem. We'll squeeze one last question in from Felix Henriksson from Nordea.

Felix, please go ahead. Go ahead, Felix.

Can you hear me now?

Yes, go ahead.

Great, perfect. Thanks for squeezing me in.

In the report, you say that the IP Networks product mix had an adverse impact on the NI gross margin. Was there something specific to the quarter or does this sort of imply that the margin profile in data center switching products at this scale is sort of dilutive to your NI gross margins?

Thank you.

I think this is largely tied to what we said at the CMD. We'd see some gross margin headwinds as we ramped products in this space, and this is what we're seeing.

What I'm focused on is the business is fundamentally accretive to gross profit and ultimately to our operating margins. Then as we talked about earlier in one of the answering one of the questions, making sure we're streamlining the company and driving efficiency so that we unlock operating leverage, and that's our focus.

Obviously, we've got to show that, but when I think about where Marco and I are focused, we're very focused on that side right now. I think you'll see the margin as we mature in this space continue to improve.

Thanks, Felix. Ladies and gentlemen, that concludes today's call.

I would like to remind you that during the call today, we have made a number of forward-looking statements that involve risks and uncertainties. Actual results may therefore differ materially from the results currently expected.

Factors that could cause such differences can be both external as well as internal operating factors. We have identified such risks in the risk factor section of our annual report on Form 20-F, which is available on our Investor Relations website.

Thank you for joining us today.