Napatech A/S

Good morning. I'm Klaus Skovrup, CFO of Napatech.

I am pleased to welcome you all to Napatech's presentation for the first quarter of 2026. Joining me today is our CEO, Kartik Srinivasan.

Our first quarter 2026 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 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 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.

Thank you, Klaus, and hello, everyone. Let me start with a brief summary of the quarter.

We saw continued strengthening in our financial performance and early signs of demand recovery in the core infrastructure market, supported by disciplined execution across the business. At the same time, we are seeing accelerating momentum in our design win pipeline across both core and AI infrastructures.

Importantly, this pipeline is increasingly progressing towards production, which we expect to translate into revenue over time. Finally, our product positioning remains highly differentiated.

As AI workloads scale, the network has emerged as a critical bottleneck and our deterministic programmable architecture is well aligned with these evolving requirements. Overall, the quarter reflects improving fundamentals, building momentum and a clear positioning for the next phase of growth.

Turning into our financial performance for the quarter. We delivered revenue of $5.7 million, representing 69% year-over-year growth, primarily driven by our improved demand in our core infrastructure business.

Gross margins remained strong at 70%, reflecting a favorable product mix and continued discipline in execution. We're also seeing improvement in revenue trends, indicating early signs of recovery in our core infrastructure markets.

With all this, while our guidance for 2026 remains unchanged, we continue to focus on consistent execution and converting pipeline into revenue over the course of the year. Turning to business momentum.

On the core infrastructure side, we saw solid activity in the quarter with 5 new design wins, continued pipeline expansion across verticals and new customer engagements. We also converted a key design win in the financial infrastructure.

This is a production-oriented engagement with a multiyear opportunity. And importantly, we view this as a repeatable use case across similar customers.

I will go into a bit more detail on this space in my next slide. On the AI and infrastructure side, we continue to make steady and tangible progress.

Our technical deliverables are on track and validation and testing activities are progressing as planned. At the same time, we are seeing continued collaboration as we advance overall solution readiness towards production.

In parallel, our engagement with a Tier 1 server OEM continues to progress with use cases defined, product deliverables aligned and commercial discussions underway. Overall, we are seeing strong execution in core infrastructure alongside meaningful progress in AI as both areas contribute meaningfully to our growth trajectory.

I'll now take a moment to highlight one of our core infrastructure verticals, financial trading networks. These are mission-critical, latency-sensitive environments where performance is defined not just by speed, but by consistency and determinism.

Typical applications in this space include real-time market data capture and normalization, feed handling, trading signal generation and order execution, where even microseconds of variation can impact outcomes. Our customers in this segment include global banks, hedge funds, proprietary trading firms and exchanges, all operating highly performance-sensitive infrastructure.

In these environments, the network sits directly in the critical path and increasingly becomes the limiting factor for performance. This is where Napatech's architecture is well aligned, enabling deterministic ultra-low latency processing with high reliability.

Importantly, this is a repeatable use case with deployments across leading financial institutions and clear expansion potential over time. Let me now turn to AI infrastructure and how Napatech's role in this space is becoming increasingly critical.

As AI workloads scale, performance is increasingly constrained by the network rather than compute. Moving data efficiently between AI compute, memory and storage has become a critical challenge.

Importantly, we view this not as a linear or evolutionary shift, but as a more fundamental change in how compute, networking and memory interact, requiring a different architecture to scale efficiently, both from a performance and energy standpoint. Traditional networking introduces variability, congestion and CPU overhead, which limits overall system efficiency and utilization.

What we enable is a fundamentally different approach, deterministic programmable networking that sits directly in the data path. This allows for consistent low latency movement, improved utilization of compute resources and more efficient scaling of AI workloads.

In practical terms, this applies across applications such as distributed inference pipelines, data preprocessing and storage access used by hyperscalers and enterprise AI deployments. Overall, we see this as a structural shift in the market where networking becomes a key lever for performance and efficiency and where our architecture is well aligned.

Before I hand it over, just to summarize, we are seeing strengthening financial performance, solid momentum in our core infrastructure business and continued progress in AI as we position for the next phase of growth. With that, I'll turn it over to Klaus to walk through the financials in more detail and provide an update on our outlook.

Thank you, Kartik. We entered the year with a strong revenue in Q1 of $5.7 million, up 69% compared to Q1 last year.

And as Kartik mentioned, the performance reflects continued customer engagement across our core infrastructure segment solutions. Our gross margin in Q1 was 70%, in line with last year.

Our staff costs and other external costs in Q1 amounted to DKK 44.5 million, down 9% compared to Q1 2025, mainly due to reduced cost of subcontractors and personnel during 2025 to balance cost to the revenue. In Q1, we capitalized DKK 810,000 compared to DKK 3.1 million in Q1 2025.

You can also read that our EBITDA in Q1 2026 amounted to a negative amount of DKK 18 million, which is an improvement of DKK 11 million compared to Q1 last year. Free cash flow in Q1 was negative of DKK 5.8 million following the negative EBITDA, which was partly covered by an improvement in our working capital.

We still have more than DKK 120 million in available cash. While we still have focus on reducing inventory, we are increasingly also preordering to make sure we can meet customer demand, especially now where we do see increased lead times for range of components.

The positive development in receivables was driven by customers paying their invoices from December 2025 during Q1 '26. Net working capital at the end of Q1 was DKK 73 million, a reduction of DKK 8 million compared to Q4 '25.

Net cash flow from financing activities for Q1 was negative DKK 23 million as we did not draw on our credit facilities by the end of Q1. We continue to manage the business with a strong focus on cost discipline and cash preservation.

Our guidance for the full year 2026 is unchanged compared to our latest reporting, as Kartik mentioned. So here, we are guiding units to be sold to be expected between 8,700 and 10,700.

We expect a revenue between $32 million and $38 million, which corresponds to around DKK 200 million to DKK 240 million, ending in the middle of the range would be equal to a growth of more than 50% compared to 2025. Our gross margin interval is expected to end between 60% and 70%, and we expect staff expenses and other external costs to end in the range of DKK 170 million to DKK 180 million.

Staff costs transferred to capitalized development costs are expected to be DKK 5 million to DKK 8 million in 2026. 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. And 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.

Can you just give some more color on how you are progressing both with the Tier 1 server OEM and with the d-Matrix relationships on the server OEM, perhaps a bit more flavor on where you are in terms of the steps towards commercialization and how you think about timing there? And then in general, both, especially on d-Matrix, whether you feel that the inventory that you added in the quarter is kind of sufficient to prep you for the ramp with d-Matrix or if we should see a proper increase in Q2?

Thank you, Christoffer for that question. So I'll kind of break this into 2 parts.

One is on the AI inference customer. We are progressing from our technical deliverable standpoint as well as commercial deliverable standpoint on track.

We remain convinced and excited about that opportunity. And again, the variable here is just the timing of this.

To your question about are we changing our supply or planning in terms of how we want to fulfill it. We're sticking to what our guidance is there.

We had already factored in orders coming in from them. And so there's no change to that plan as far as that AI infrastructure customer is concerned.

And as far as the Tier 1 server OEM, Christoffer, that's progressing well on plan. We have alignment on the statement of work.

We have alignment on the features, performance, cost, schedule, et cetera. So things are looking good there.

Again, the same kind of response applies. We are delivering to our requirements.

When that converts into some sort of meaningful thing remains to be seen. But we have some amount of orders in 2026 baked in from the Tier 1 server OEM as well, which we are not changing right now.

Just on the 5 design wins, could you perhaps give us a bit more flavor on those in terms of potential and also timeline? Are they impacting '26 by any means?

And also on the recent design win within the financials, when do you expect to see it convert into the revenue that you have guided for? Is that in Q2 or in Q4?

Or when is that going to happen?

Yes. Thank you for the question.

The design win momentum that we speak about, both of those are pertaining to the core infrastructure the 5 new design wins that we had, each of them will track a different kind of path towards production. And some of these may actually impact 2026, but again, no change to our plan there.

But these tend to be a bit drawn in how they convert from a design win into full production. It can be anywhere between 6 and 9 months for an existing product, which is why the core infrastructure market is so lucrative for us.

We do turnkey solutions in that space to an existing well-known robust customer base. On the other side of the key design win that we announced -- on the financial side, we are expecting the first $1.5 million that we spoke about, that is expected within the next couple of months.

So that will definitely hit our 2026 revenue, and that's kind of part of our plan. And the rest of the opportunity will kind of extend into '27 and beyond there.

So that's kind of how the timing of the financial institution revenue looks like.

What's the financial design win baked into your original guidance?

Yes. So the way this works is we do have a lot of proof-of-concept designs that we do on an annual basis.

And at any given point, each one of these or a few of these can actually convert into a design win like this for us. So we've done data analytics based on how we are engaging with these customers or segments, and we baked that into our assessment for 2026.

So this was not something that came as a complete surprise. This is already baked into what we were expecting.

Congrats on the strong growth in Q1. So I guess this kind of reflects the good momentum in your traditional SmartNIC business.

Can you say kind of what kind of verticals, what kind of use cases are you seeing strong growth? And how you kind of expect kind of the traditional SmartNIC business to move throughout the year?

Should we anticipate kind of normal seasonality from here that it strengthens in the second half? Or is kind of some kind of a large one-off contract or something in Q1, so we shouldn't kind of fully extrapolate that?

Yes. Fantastic question.

Thank you for asking that. So we do -- there is a reason why we kind of placed the definitions of our market segments as core infrastructure and AI infrastructure.

Within the core infrastructure, which is where we are seeing early signs of kind of demand recovery, that breaks down for us into at least 4 big verticals that is fintech, telco, cybersecurity and network packet monitoring. Each one of those segments represents individual growth areas for us.

Like we announced in the fintech space, we converted a design win into revenue. But we also expect alongside the cybersecurity, packet monitoring and telco businesses are showing signs of early recovery as well.

So we're expecting that to play a part in our 2026 performance.

And can you say something about in terms of new design wins in the AI infrastructure area? Are you working on many leads there?

What are you seeing in that area in terms of new potential design wins?

Yes, there is a lot of engagement there. That is actually the more fascinating place for how fast that space is evolving.

One thing I can definitely tell you with a lot of confidence is the AI infrastructure is no longer a compute problem. It is an efficiency problem, which immediately translates to 2 components of the data center that come into question.

One is networking and the other is memory. And you will see that, that's why there is a lot of requirements, a lot of specifications, a lot of consortium being formed around what to do in the space of networking.

Big hyperscalers are putting out specs, consortiums are putting out specs. And this is an exciting time for a company like Napatech because we are fundamentally based on programmable networking.

Our architecture, our products are built to deliver to an evolving paradigm, which is what the AI infrastructure market is going through right now. So I am very confident and excited about this space, and there's a lot of activity that we are currently in, and I'll be the first one to come here and tell you as they convert into some sort of design wins.

Thank you. And we do have some written questions here.

I'll just see whether I can group them a little bit. So there's one here from [indiscernible].

A few questions. What is your visibility on H2 AI orders?

How will ramp look like? And maybe we'll just start with that and then we'll take the rest of the question.

I think you spoke a little bit into it already, Kartik.

Yes. So the second half of our 2026 is going to be exciting as well because that's when we start seeing some of our AI infrastructure revenue starting to materialize.

I think we had mentioned that a good chunk of the 2026 revenue for us will still be based on core infrastructure and the AI infrastructure will probably be about 20%, and that's kind of still the mix that we're expecting for 2026.

And then regarding the Tier 1 OEM, have you lost any opportunities?

No. Actually, we have not.

In fact, we are in a very favorable position within the Tier 1 OEM. I remember using an expression the last time I was here called the hub-and-spoke model, which basically translates to we are very much engaged with the technology team, which is the hub, and then there is a few spokes that lead to the product teams.

That model is still strong for us, and we are excited about delivering what we are calling the next-generation data reduction technology, which is extremely important in the AI infrastructure world as you're loading in huge large language models, compression and deduplication becomes a very important aspect of how data gets rolled out. And that's exactly the critical application or use case that we are delivering in this proof of concept.

And then maybe just jumping to [indiscernible], who has a question here. How is the progress in proof of concept with the Tier 1?

That's what you just explained. And when can we expect to be part of their sales in their server sales?

Yes. So the conversion of these designs into revenue and productization, that's kind of the variable in play here.

We do have some level of early orders baked into our 2026 revenue like we discussed. But when that converts into material production and revenue, that remains to be seen.

Yes. And then a follow-up question for Lars Knutsen (sic) here.

Are there other opportunities arising with other players within AI infrastructure?

Yes. So that is the engagement or the set of engagements that I've been talking about.

We do have to the capability that we have as a company to deliver, we have been selective on how we engage because at the end of the day, we have to prioritize how we deliver and make sure that our deliverable is robust. But yes, there is a set of engagements that we have on the AI infrastructure, which follows our design win pipeline, 4-stage pipeline routine.

Then there's a question here from Ola Millingstad (sic). You're right that first orders from AI infrastructure will come in H2.

Could you be specific how many dollars of your '26 guidance of USD 30 million to USD 37 million assume AI infrastructure deliveries? And if H2 AI orders come in as, how does that change the full year picture?

Yes. So our 2026 revenue profile will still be significantly biased towards the core infrastructure.

I think we -- last time we were here, we said about 20% of that is going to be coming in from AI. That's what the numbers look like.

And of course, if the AI orders don't materialize, then that's what the impact is going to be.

And let me just see other related to this, I think that is in earlier presentation, this [indiscernible] in earlier presentation, it has been said just 20% of the forecast from d-Matrix is taken into Napa's forecast for '26. Is this still valid?

Yes. yes.

So I think that observation of yours is still what we are tracking for 2026.

I think the point here is that d-Matrix may have a higher forecast and then we have only taken 20% of that in. So I think the point is that currently, our guidance that's based on our expectations of what orders are coming from d-Matrix.

Right. Right, correct.

Yes. Then there's one here from [indiscernible] again.

d-Matrix acquired GigaIO to build a complete rack scale system. How has that changed your role at d-Matrix?

Are you still the scale-out NIC in the SquadRack reference architecture? Or is there a risk they switch to a different supplier as they build out their own stack?

A good question, and I'm sure a lot of people have the same thing, and I'm so a lot, I'm glad that you asked it. Our relationship with them remains as strong as ever.

In fact, it's even getting stronger because of the requirements that are unfolding for us as they themselves are evolving in this landscape very rapidly. The requirements are coming in towards production, and we have completed the first stage, and we're already embarking on the next-generation kind of engagement to see what that looks like.

The Giga -- and of course, a lot of these questions are for the d-Matrix to answer. But the acquisition that they made of the GigaIO data center assets that definitely enables them to build more of a rack scale architecture of which we are singularly the component delivering scale-out networking.

And then there's a question here from Lisa [indiscernible]. You cite commercial discussions underway on the Tier 1 server OEM design win.

Does that mean that proof of concept has been achieved? Where are you currently in the qualification process?

Right. As part of the design win engagement that we are having with the Tier 1, starting commercial discussions typically indicates that there is activity across the 3 groups that make decisions at these kind of places.

So we have alignment with the executives. We have alignment with product management and engineering, and now we are working on alignment with commercial.

And once all of those move forward, that converts into product and then converts into revenue. So moving into the commercial stage or activating the commercial stage is a strong indication that things are making good progress towards production.

Yes. And also from Lisa here, can you also give some color on expected revenue ramp-up for the remainder of the year, given that Q1 was seasonally very strong.

Yes. So there is definitely seasonality that we will still continue tracking towards -- throughout our 2026 revenue.

And because our 2026 revenue is still heavily based on core infrastructure, the seasonality will mimic prior year seasonality, not necessarily in the absolute dollars, but at least in direction. So you can expect some of that to be similar in 2026.

Yes. And then a question here from Terry Jensen.

Why is revenue for Q1 lower than revenue in Q4? And I think that was what you just explained, Kartik, also that we do see seasonality.

And usually, Q1 is our weakest quarter, whereas Q4 is the strongest. And then there's the last question here from [indiscernible].

What's the time line from your order cart from you order carts to have it in stock? Any limitations?

Yes. So again, very good question across the overall supply chain management and planning there.

So our demand planning cycle or at least process is no longer limited by demand. It's actually just what we have to -- how we manage our supply.

The lead times across critical components that we put in our adapters and solutions, including the likes of FPGAs and the memory components are seeing longer lead times and, of course, volatility in availability. But the good news is that our engagement with our customers allows us to have a solid level of visibility into how we plan our quarterly delivery.

And we have put some mitigation factors in place as well by some level of prepurchasing of these parts, both across FPGAs and memory. So this allows us to deliver our products to plan to the '26 revenue as well as towards what the customer requirements are in the quarterly distribution of our products.

Good. I think that summed up the last question we had in written form as in the Q&A.

Good. And I don't think there's more coming in here.

Operator, is there more questions on the line?

Thank you. Then I want to thank everyone for listening in.

Enjoy the day, everyone. Thank you.

Thank you.