Operator: Greetings, ladies and gentlemen, and welcome to the Napatech H1 2026 Report Call. Please note that this call is being recorded. [Operator Instructions] It is now my pleasure to introduce Klaus Skovrup, CFO. Please go ahead.
Klaus Skovrup: Good morning. I'm Klaus Skovrup, CFO of Napatech. I'm pleased to welcome you all to Napatech's presentation for the second quarter and first half year of '26. Joining me today is our CEO, Kartik Srinivasan. Our first half year '26 report was released earlier this morning on the Oslo Stock Exchange and is also available on the Investor Relations section of the Napatech website. For your information, a recording of this webcast will be available later today. There will be a question-and-answer session following the presentation during and after these prepared remarks, you may submit your questions via text on the webcast page or we can take your questions on the phone. If you would like to ask a question, please follow the instructions on this slide. Please note that this presentation contains forward-looking statements that are subject to risks and uncertainties. Our actual results may differ from those discussed in the forward-looking statements. For further information on risk factors, please see company announcement and the slides prepared for this presentation. With that, over to you, Kartik.
Kartik Srinivasan: Thank you very much, Klaus, and hello, everybody. Let me spend the next few minutes talking about the business and the momentum we are seeing across both core infrastructure and AI infrastructure. As we look at Napatech today, there are 4 things I want investors to take away. First, our core infrastructure business continues to provide a strong and profitable foundation for the company. Second, AI infrastructure is increasingly becoming a production story rather than simply a design win story. We have moved from engineering engagement to design win to initial production and now to a follow-on production order. That progression is an important proof point for us. As volumes scale, we're building the operational capabilities and supply chain needed to support that growth. Third, we're deliberately broadening the customer and partner ecosystem around both businesses. That gives us more routes to market and over time, should help AI infrastructure develop into a portfolio of opportunities rather than a single customer story. Finally, as inference becomes increasingly heterogeneous and distributed, networking requirements are evolving alongside the computer architecture themselves. That is precisely where programmability matters because the infrastructure can adapt as those architectures evolve. Put simply, core infrastructure is our strength today, while AI infrastructure represents a significant growth opportunity for tomorrow. You can see that momentum in our first half results. In Q2, we generated revenue of USD 7.4 million, an increase of 55% year-over-year with a gross margin of 67%. For the first half, revenue was $13.1 million, up 61% year-over-year, while unit volumes increased 20%. The difference between revenue and unit growth also reflects an important mix shift in our business. We are seeing demand move towards higher speed, higher-value products, which carry higher average selling prices. As that mix continues to evolve, revenue growth will not necessarily translate proportionately into unit growth. This is a trend we highlighted during our full year 2025 earnings, and we are continuing to see it play out in 2026. Importantly, this growth has been driven primarily by our core infrastructure business, while we also have begun 2 qualification orders for our AI infrastructure. We are delivering strong growth from the established business while at the same time advancing our next growth engine. Based on our performance through the first half and our current visibility, our full year 2026 revenue guidance remains unchanged. Here, we show the role that each of our 2 businesses plays in the Napatech story. Core infrastructure remains the profitable foundation of the company. First half revenue grew 65% year-over-year to $12.8 million at approximately 70% gross margin. We also added 12 design wins in this segment, which gives us additional opportunities for future revenue growth. At the same time, we're building AI infrastructure as our next growth engine. The revenue contribution today is still relatively small as these programs progress through engineering, validation and commercialization. But importantly, we are beginning to see that activity translate into production orders and higher volume deployments. As you would expect with programs of this scale and complexity, individual opportunities will progress at different rates. Our engagement with our Tier 1 OEM, for example, has moved more slowly than we originally anticipated, largely due to organizational and management changes at the customer. The technical engagement remains active, and we continue to see significant strategic value in the opportunities we are pursuing together. More broadly, our AI infrastructure pipeline continues to grow, providing multiple parts to future revenue. As these opportunities move toward higher volume production, we're building the operational, manufacturing and supply chain capabilities required to support that scale. We also expect to continue investing as the business grows. But importantly, we are building our AI technology for reuse and our organization for operating leverage. As AI infrastructure revenue scales, we do not expect operating expenses to increase proportionately with that revenue. Our objective is not simply to grow revenue, but to translate that growth into increasing operating leverage over time. Before I get into the numbers on this slide, I want to point out this is a new view for us. We haven't historically presented our revenue geographically in this way. But as Napatech grows and our strategy evolves, we believe it's useful to give investors greater visibility into where our growth is coming from and where we are investing for the future. We see this as an additional strategic lens into the business that will help investors better understand how Napatech is developing over time. In the Americas, first half revenue grew 27% to $7.3 million. This is also where we are making a significant portion of our strategic investment in AI infrastructure, and we are seeing strong engagement with our customers across both core and AI infrastructure. In the Rest of the World, revenue grew 141% to $5.9 million, driven by strong execution across our core infrastructure customers. This geographic view gives you another perspective on our strategy, strong growth across our established markets alongside targeted investment in the areas where we see the greatest opportunity for future expansion. Let me finish my section by stepping back and talking about why we believe AI infrastructure represents such an important opportunity for Napatech. AI inference is changing rapidly. The industry is moving beyond architectures built around a single type of processor. Increasingly, AI systems are becoming heterogeneous, combining CPUs, GPUs and specialized accelerators, each optimized for different parts of the workload. We are seeing this direction validated across the industry, including by pioneers such as NVIDIA and other leading AI infrastructure companies. Importantly, we believe the shift towards heterogeneous compute plays directly to the value proposition of Napatech's Adaptive AI NICs. In this new world, established leaders will increasingly coexist with a new generation of specialized AI companies, each bringing different compute architectures and capabilities. The more diverse that compute environment becomes, the more important and more complex the infrastructure connecting it becomes. The challenge is no longer simply about how much compute you have. It is also how efficiently you can connect that compute and move data between it. That has a direct impact on performance, utilization and ultimately, the economics of AI inference. We are seeing these architectures evolve rapidly across the industry, including an increasing focus on high-performance networking technologies such as RDMA. This is where our programmability becomes particularly important. Our Adaptive AI NICs are built on programmable architecture. As processors, accelerators, networking protocols and workloads evolve, our technology can adapt with them. Our goal is simple: help our customers move data more efficiently, keep their expensive AI compute better utilized and ultimately improve the economics of AI inference. As AI infrastructure becomes more heterogeneous and more complex, we believe the value of that programmability increases, further differentiating Napatech and strengthening the moat around our Adaptive NIC architecture. That combination of adaptability and programmability is at the heart of AI infrastructure business we are building. We believe this growing relevance, combined with the differentiation of our programmable architecture is an important part of the long-term value creation potential for Napatech. With that, I'll hand it back to Klaus to take you through our financial performance and outlook in more detail.
Klaus Skovrup: Thank you, Kartik. We delivered strong revenue growth and improved operating performance in Q2 while maintaining disciplined cost control. As Kartik mentioned earlier, our revenue in Q2 was $7.4 million, up 55% compared to Q2 last year. In Danish kroner, revenue increased 52% to DKK 48.1 million. Our core infrastructure segment is driving profitability in Q2, while the activity in the AI infrastructure was driven by early customer qualification orders coming at low margins due to the limited quantities produced. When production increases, the marginal unit cost will decrease and margins improve. Our gross margin was 67.3%, slightly above Q2 last year and within our guidance range. For the first half year, revenue was DKK 84.7 million, up 53% compared to last year, and gross margin was 68.3%, essentially in line with last year. Staff costs and other external costs were DKK 41.2 million in Q2, down 3.4% compared to Q2 '25, reflecting continued cost discipline. EBITDA improved significantly to a negative DKK 6.6 million compared to negative DKK 19.6 million in Q2 last year. While we are still investing in the business, the operating leverage is clearly improving as revenue scales. Free cash flow in Q2 was negative DKK 35.3 million, mainly driven by working capital movements during the quarter. Net cash flow from operating activities was negative DKK 32.1 million in Q2. The main driver was a DKK 26.1 million negative working capital adjustment, primarily reflecting backloaded invoicing in Q2 and a ramp-up of inventories to support expected customer demand. For the first half year, net cash flow from operating activities was negative DKK 36.9 million compared with negative DKK 44.3 million in the first half year of '25. Free cash flow was negative DKK 41.1 million, which is an improvement compared with negative DKK 50.6 million in the first half year of last year. Net working capital increased from around DKK 73 million at the end of Q1 to around DKK 99 million at the end of Q2 due to the increase in inventories and receivables, as mentioned. Cash and cash equivalents were DKK 62.7 million at the end of Q2 and including undrawn committed credit facilities of DKK 21.6 million, total available liquidity was DKK 84.3 million. The year-to-date cash development reflects both the improved operating performance and the higher working capital tied to inventory and customer timing. We continue to manage the business with a strong focus on cost discipline, cash preservation and ensuring that working capital supports expected production ramps and customer demand. Our revenue guidance for the full year 2026 is unchanged at DKK 200 million to DKK 240 million, corresponding to approximately USD 32 million to USD 38 million. As Kartik mentioned, we are seeing demand move towards higher speed, higher-value products, which carry higher average selling prices. As that mix continues to evolve, revenue growth will not necessarily translate proportionately into unit growth. Furthermore, we are expecting lower than originally anticipated units sold within AI infrastructure in 2026. This leads us to a lower unit guidance for expected units sold in '26 to the range of 7,700 to 9,700. We maintain our gross margin guidance of 60% to 70%, though you should note that in the next quarters, we expect gross margins to be lower following a higher share of AI infrastructure revenue coming at a lower margin and increased component prices here under especially memory. Staff expenses and other external costs are expected to be DKK 170 million to DKK 180 million, while staff costs transferred to capitalized development costs are expected to be DKK 5 million to DKK 8 million, both are unchanged. At the midpoint of our guided range, EBITDA would be negative at around DKK 25 million for the full year. As always, our outlook remains subject to normal risks, including currency movements, market uncertainty, trade barriers and supply chain volatility, which we continue to monitor closely. As we wrap up today's presentation, we would like to invite you to visit Napatech at one of these upcoming events. Our full year event plan is shown online at the link provided. If you happen to be in one of these great cities during the coming period, we would love to meet you in person. With that, we are now ready for the Q&A. Operator, we are now ready to take the first question.
Operator: [Operator Instructions] Your first question comes from the line of Christoffer Bjornsen of DNB Carnegie.
Christoffer Bjørnsen: First of all, on the Core Infra segment, it seems to be progressing well. I would just love to hear Kartik's reflections on that business after been in the company for a while. I think previously, at least the previous management was obviously more excited about the AI Infra segment, but now it seems the Core Infra is also booming. I think some of the traditional customers within that space, for instance, within trading and financial services have kind of surfaced as some of the big buyers of AI compute in the billions of dollars from the likes of neocloud and new acceleration companies. So do you kind of -- Kartik, you see an opportunity for you guys to be attached to those kind of ramps within the more traditional space? Or is that something you're not necessarily playing in? Just your reflections on the Core Infra segment going forward, if it's exciting or not will be appreciated.
Kartik Srinivasan: Christoffer, thank you very much, as always, for your question. We do feel the core infrastructure market has got a huge SAM. As part of that SAM, we, as Napatech, were partaking in that SAM in a fairly limited capacity until now with a single product that we call Capture. In that space, the market signals are positive, and we see a resurgence in demand in there as reflected in our earnings. As the core infrastructure market is recovering and the demand signals are growing, we remain excited about the core infrastructure space as well. Of course, the AI infrastructure market doesn't need any explanation. It's very, very exciting for everybody. It's dynamic in nature, and we are very well placed as a company with our programmability to participate in that success. But at the same time, our focus remains unchanged on the core infrastructure side as well. That's a market that we understand that's a go-to-market that we understand the customers and the partners are something that we have worked with for many years. Christoffer, that is the core infrastructure market remains a focus area for us over and above what we're doing in the AI infrastructure space.
Christoffer Bjørnsen: Just double question on the AI inference and AI infrastructure space. So I think, first of all, on the current lead customers that have given you these initial orders. Maybe you can't say too much. But I think last quarter, you gave an update on how you're developing or progressing with their next-generation efforts. They have new products coming out, new architecture. Just how you see yourself being positioned to have a role there as well as they move to a completely different architecture on the main kind of compute side? And then secondly, on AI infra, just any update and more granularity on other opportunities within AI -- in just how those are progressing? What kind of roles are you seeing there? And kind of if you see any timing there you could comment on beyond the metrics, that would be appreciated as well.
Kartik Srinivasan: Awesome. Sounds good. Yes. Our leading AI infrastructure customer, as you know, there's a lot of industry information available, not just on how their individual next-generation architecture is evolving, but also in the last 6 months or so, heterogeneous compute has become pretty much common place. Anytime you hear AI inference, you hear heterogeneous compute in there. The role that Napatech plays across both of these individually within our leading customer and their next generation as well as how we participate in the heterogeneous, they are kind of related, Christopher, in the sense that we went from a dedicated kind of offering for their first gen to a more industry RDMA-specific offering that we are working on, and that is an engagement that we are currently doing with our customers, both across the lead customer as well as the industry. The industry is gravitating towards RDMA, and you can see in any journal that now there are 4 or 5 different kinds of RDMA technology. There is RoCE and then there is UEC, and then there is MRC and then most recently, Meta announced MetaRoCE. That's a very dynamically changing landscape. The underlying programmable architecture that Napatech has now becomes an extremely important weapon for us in playing in this field as this RDMA is changing so dynamically, and it becomes a differentiation for us.
Christoffer Bjørnsen: That's super helpful. And then on the other opportunities, anything there, anything material you can disclose in terms of incoming or outgoing calls or opportunities?
Kartik Srinivasan: Yes. So yes, Christoffer, you have to assume that our pipeline is something that we are actively working on. But right now, it's a little too soon for me to come in and provide you with any sort of guidance in terms of names or numbers. But rest assured that as soon as I am able to do that, you will know.
Operator: [Operator Instructions] Your next question comes from the line of Christopher Bjornsen of DNB.
Kartik Srinivasan: You went back in to the line.
Christoffer Bjørnsen: That's great. I just have to ask on the cash flow and the balance sheet kind of a housekeeping question. You reflected somewhat on it, but just how do you see the runway and how do you see the working capital requirements and so on into the second half and beyond? I see your current cash -- negative cash flow is driven by working capital to a large extent. You seem like you could be on like an operating free cash flow basis into positive territory in the second half at some point. So just some reflections on how you see that progressing through the second half of the year?
Klaus Skovrup: Yes. Thank you, Christoffer. You're absolutely right. The reason why we had this cash burn in Q2 was that we built for inventory to meet the demand here in the second half year and then Q2 was also backloaded. We did a lot of the invoicing in June, and thereby, we'll first receive the payments for those in July or August. While we go into the second half year, our expectations currently are that we will exceed DKK 10 million in revenues in the coming quarters. Thereby, we should be close to cash flow neutral and hopefully even cash flow positive also in the next 2 quarters.
Operator: Your next question comes from the line of Oystein Lodgaard of ABG.
Øystein Lodgaard: I have starting with a question on the Tier 1 server manufacturer. You're saying the ramp-up there goes a bit slower. I know it's very difficult to comment on specific customers. But are there any kind of things you can share about the progress with that customer? Are you still working on several projects with that customer? You have stated previously 6 ongoing projects. And do you still see kind of the opportunity there as big as it has been previously?
Kartik Srinivasan: Oystein, thank you for your question. Absolutely. undoubtedly, the engagement at the technical level remains extremely strong. I think as I mentioned the last time, these big companies have a very disciplined multi-phased approach going from qualification to production. These big companies, unfortunately, also face a lot of management changes, and that's what we are in the middle of. This is nothing more than just a little bit of a delay in the progress, but it doesn't take anything away from the activity or the size and kind of scope of the overall effort.
Øystein Lodgaard: That's very good to hear. And on the core infrastructure segment, it's good to see that, that is also performing now very well, growing strongly there in the first half of the year. If we look into 2027, what's kind of your visibility for the core infrastructure segment in '27? Do you have new design wins to support continued growth in '27? Should we kind of expect it to continue to grow at the rates we are currently seeing? Or is this kind of a bit of an extraordinary growth this year in that segment?
Kartik Srinivasan: Yes, good question. How much the core infrastructure grows by in '27, a little too early for me to comment on. I'm pretty sure over the next quarter or maybe 1.5 quarters, I can get you better visibility into the size and the rate of growth, Oystein. But I can tell you, directionally, we are expecting growth, of course, in our core infrastructure and needless to say, in the AI infrastructure in a meaningful way next year. We are investing and we will continue to invest in both of these segments because we fully expect both of them to grow. How much they grow by, Oystein, we'll come back and give you the number.
Klaus Skovrup: I think there's one more question in the line, right, operator?
Operator: Yes. Our next question comes from the line of [ Lars Schmitz ], a private investor.
Unknown Attendee: So I think the CEO of d-Matrix, he gave an interview, I think, a few days ago, where he talked about that they're deploying 1 to 2 megawatts of compute this year and that, that will grow to 30 to 50 megawatts of compute in '27. I mean I'm a bit puzzled how to read that because I just look at it, it's like a 25x in terms of compute, i.e., also cost. Is that also something that you're recognizing when you have discussions with d-Matrix?
Kartik Srinivasan: Yes. Thank you for the question, Lars. The conversion of the wattage of these data centers that the compute architecture teams plan and how that could get converted actually into the networking component, which is what Napatech provides, the math or the arithmetic there, Lars, is not that easy. But what we have visibility to is a direct forecast and POs that we get from our customer. That's what we plan based on and then we, of course, have some level of buffer so that we can plan our supply chain. We're obviously also monitoring what some of these public statements are, and we'll start building our models over time to see how these exact wattage of the data center converts to forecasting units and potentially revenue for us going forward. But still, again, this is such a dynamic business right now. The direct conversion of a component such as what Napatech provides to this landscape and getting that equated to the bigger picture of what the megawatts or the gigawatts conversion is, is a little difficult today.
Unknown Attendee: And second, just on the order pattern from d-Matrix, do you have any insights into the 2 orders that you have received, are they for different customers? Is it one customer scaling? And I mean, we are very early in this journey, but the first order was for like Q3, Q4. The second order was for Q1. Will the next order, will that be for Q1 '27? Or will it be for Q2, '27? How do you see the order pattern and what insights do you have into which customers they are deploying at?
Kartik Srinivasan: Yes. Lars, another question where the answer is tough to predict. All we know is the first order that came in, that's for delivery in '26 and the second order that we announced last month was for starting in Q1 '27. That's pretty much the information that we've been directly provided. Everything else that -- all the other questions that you asked, I don't have the answers to.
Operator: I'd now like to hand the call back to the Napatech team to address the web questions.
Klaus Skovrup: Thank you. We got a question from Anders Knudsen. Actually, a couple of questions. So I'll just read out loud here. Congrats on the strong revenue growth. Could you add some more detail on how memory cost increase impacts the gross margin? How big of a bill of material is memory? I think the answer to that one, Anders, is that it's a quite significant part of the bill of material. Maybe let me just give a quick example, hypothetical example, but it gives some flavor to it. So let's play that we sell a unit at $5,000 we have a margin of 70% on that $5,000, so $3,500 in gross profit margin, thereby the COGS is $1,500. If the COGS suddenly increases to $3,000 due to increased memory costs, we keep the price at $5,000, then our margins would decrease to around 40%. If we wanted to keep the margin of 70%, then we needed to sell the product instead of for $5,000 at $10,000, so double up the price. That's, of course, where it gets a little bit difficult because customers, they also know that memory prices is not increasing that much. So they need to pay double the price for the product they have before. But what we are doing is, of course, to get the highest price we can to get the best possible gross profit margin for Napatech. Then there's a second question here. Given the acceleration in design win, how big is your pipeline nowadays, fine? And maybe, Kartik, do you want to answer that one?
Kartik Srinivasan: Yes. The pipeline nowadays for both on the core infrastructure side that we announced the 12 new design wins as well as on the AI side, it's a growing number for us. For 2026, the design wins that we announced this year, obviously, will not have an impact. But the prior announcements that we made for design wins, those will obviously come into play for 2026. The question is what are the core levers to hit the top end of the guidance? It really is -- the demand signals are very clear for us. It's now just a matter of how we manage our deliveries, how we manage the pricing. And the guidance that we have provided, I think that's -- we're still holding on to that guidance. We'll manage it in that range.
Klaus Skovrup: Yes. We are making sure that we -- with the product mix, we expect that we have enough in our inventories to meet the demand to end in the high end of the guidance.
Kartik Srinivasan: Yes. One thing I will reiterate just because the question is in that space. I know we announced that the unit count forecast or the guidance has come down, but I do want to emphasize that this is coming at a -- the lesser units do come at higher value, higher ASP for Napatech, which we see as positive news. Because it's now our customers are moving to a higher speed of Ethernet, higher value of our products. That kind of is the reason why despite the forecast of units going down, we are not changing the revenue guidance in here at all. I do want all of the investors to notice that this is -- the revenue is being attached to a higher value product that Napatech is shipping.
Klaus Skovrup: Good. Then there's a question from [ Tore Avolant ]. On Slide 11, sorry, you say AI infrastructure products come with a lower gross margin. Can you give us a sense of what margin we should be modeling on the production orders and whether that improves as volumes scale?
Kartik Srinivasan: Yes. The exact number of what the margin will be, we will get to know that because there is still some level of volatility on pricing owing to this, the memory changes and some of the other pricing of our components. As we scale, and we'll have more visibility, Tore, we'll come back and give you some level of guidance in here. But our objective within Napatech, of course, is to continuously improve margins across every possible way we can, and those come from different aspects of managing our supply chain well, managing our value that we add into our solution that dictates the pricing. Those will forever be continuous efforts by Napatech to improve the margin in this space. But how exactly we model with those numbers, we will get a little bit more educated over time.
Klaus Skovrup: Thank you. I think that was the last question we had in the comment space. I think, operator, you can confirm we don't have anyone on the line either, right?
Operator: We don't have any questions as of this moment from the conference line.
Klaus Skovrup: Perfect. Then I think that wraps up the Q2 webcast. Thank you, everyone, for participating, calling in and also sending out the questions. Thank you very much. Sorry, there's just come one more question in on the last minute here from Benjamin. Let's say you get an order of 20,000 units instead of 2,000 units like you have today, would you be able to deliver on this type of big orders?
Kartik Srinivasan: Thank you, Benjamin, for that fantastic question. I would love to get that order. But the short answer is yes. We are -- have been anticipating a big explosive growth in the AI infrastructure space and have been planning for our supply chain, both across our contract manufacturers as well as the different vendors in the landscape that we have in our supply chain. We are gearing up for this level of growth in this order of magnitude.
Klaus Skovrup: Maybe I just comment from a liquidity point of view, what we would like is, of course, to get these big orders, but maybe come a little bit growing so we can also follow it from a liquidity point of view with our working capital because there is, of course, some commitments we need to do there and there we need to figure out how to finance if we need to produce such a big order potentially by getting prepayments or similar. Good. I think that's the last question then. I'll just thank you, everyone, again for attending. So operator, I think we are ready to close the call.
Operator: Thank you very much. Thank you, everyone, for attending today's call. You may now disconnect. Goodbye.