RaySearch Laboratories AB (publ)

A warm welcome to Research 2025 Year End Results presentation. My name is Karolina Stramblid, and I'm Head of Investor Relations. With me today are our CEO and founder, Johan Lööf, and our CFO, Nina Grønberg, who will take you through the key highlights and financial results. After the presentation, we will open up for questions. Feel free to submit them in the Q&A chat or ask them live. With that said, Johan, over to you.

Thank you, Carolina, and welcome again, everyone. So this is the agenda for this webcast. I will start with an interaction about research. Then I will summarize Q4 and the full year. After that, Nina Granberg will talk about the financial development. Then I'll take over again and mention the dividend proposal that we have. Since there is a strong focus on AI these days, I will make a deep dive into what that means for research. And then I will just summarize the presentation and make an outlook. After that, we take Q&A. So, a few words about RaySearch. RaySearch is a pure software company, and we are dedicated to cancer treatment software. And we have four platforms, RayStation, RayCare, RayIntelligence and RayCommand. What we see in this image is a comprehensive cancer center. And our long-term goal is to support such a center with all the software that they need. So not only radiotherapy, but also support for chemotherapy, surgery, tumor board meetings, and other things. So that's the long-term vision for research. So far we have focused mainly on radiation therapy of cancer. What we see in this image is the user interface of RayStation, our treatment planning system. And it summarizes quite well what's going on in treatment planning for radiotherapy. In the upper left, upper, let's see, upper right image here, you see a machine. This particular machine happens to be an Oxray by Hitachi. And we can see in this image how the machine moves around the patient. So it rotates and it also swivels this ring around the patient. So one thing that we have to do in our treatment planning system is to model this machine so we understand how we can move and also the physics of the beams. We can calculate dose in the patient, et cetera. The next thing we need is a model of the patient. So we see the patient in the middle. upper image here and also in the other images you see different cross-section of the patient and we also see the dose distribution which is the color wash overlaid on these CT images and the idea here is that we get a high dose to the tumor which is the red color in this image and low dose to the organs at risk outside of the tumor for example you see the spinal cord in this in this sagittal view, that it has a very low dose. The lower right image shows the patient from the source. If you look at the patient from the source, this is how the patient would move. It looks like the patient rotates, but it's actually the source that rotates around the patient. So this is in summary what we are doing in treatment planning for radiotherapy. We also want our systems to absorb data as we treat the patients, so that all of our products will automatically capture the data that's being generated before the treatment starts, during the treatment, and during follow-up, what happens to the patient after the treatment. And by absorbing all of this data in the system intelligence, we can achieve clinical insights and feed information back to our systems to improve the systems. And some of those icons represent machine learning models, but it can also be other aspects, other types of clinical insights. And we use this feedback data to improve our algorithms. We have some examples of that already out in the field. We can improve the efficiency of the operation. Ultimately, we want to provide decision support so that the members of the tumor board, for example, can make the best possible choice of strategy for the patient and ultimately improve outcomes. I show this diagram just to illustrate the long-term journey in terms of revenues. The reason is I want to highlight that we shouldn't look at research revenues on a quarterly basis. If you zoom out a little bit, this one goes all the way back to 2008. we can see that there has been a steady growth of revenues year after year. The two pandemic years are an exception, and we understand why those two years were weaker. But besides those, there has been a steady growth of the company, even though you may see fluctuations between quarters. Okay, and now I will make a few comments about the last quarter and also the full year. So Q4 was a strong finish of the year. Net sales grew by 16% to 375 million Swedish crowns, which is all-time high. Adjusted for the strong currency headwinds, the growth would have been 28%. Recurring support revenue was 139 million Swedish kronor, which corresponds to 37% of the total revenues. The strong sales translated directly into improved profitability. Operating profit increased by 25% to 92 million Swedish kronor, resulting in a 24% EBIT margin. Adjusted for personal losses, the margin would have been 27%. So for the full year 2025, net sales increased by 13% to 1.34 billion Swedish grams, marking the highest annual revenue in the company's history. Organically, net sales grew by 19%. Recurring support revenue was 524 million, which corresponds to 39% of the total revenues. Operating profit was 292 million Swedish crowns for the full year and the margin 22%. If we adjust for currency effects and extraordinary items, the operating profit would have been 353 million and the margin 26%. Let me briefly highlight a few of the new orders and expanded installations we secured during the quarter. We continue to see solid momentum with strong license sales to both new and existing customers across all regions. Greater Poland Cancer Center expanded its registration installation to include proton therapy, bringing photon and proton planning together on a single platform. The University of Pennsylvania, one of the premier proton therapy institutions in the US, selected RayStation as a unified treatment planning system for proton therapy across its three clinics. Universitetsklinikum Gießen and Marburg in Germany chose to replace Philips Pinnacle, which reaches the end of life in 2027, with RayStation. We have also seen strong clinical progress during the quarter. The Royal Marston NHS Foundation Trust achieved a major milestone by performing its first online adaptive treatment on a standard electrical DYNAC using RayStation's adaptive planning module. Until now, most online adaptive treatments have been limited to specialized machines that few centers have. This achievement makes online adaptive radiotherapy accessible to far more clinics and patients. At the Southwest Florida Proton Center, the first patient treatments were delivered using RayStation and RayCare together with IBA's proton therapy system, enabling highly precise treatments including proton arc therapy. Together with the Trenton Proton Therapy Center, we performed the world's first clinical proton arc treatments in 2025. A technique that improves dose distribution by using many beam angles and optimized energy levels. This achievement was actually named one of the top 10 scientific breakthroughs of the year across the entire field of physics by Physics World. And that's something that we're very proud of. And now I will hand over to Nina, who will go through in more detail the financial development.

Thank you, Johan. Taking off from your presentation and the numbers in brief, we can conclude that it has been high interest in research solutions throughout the year, and that goes both from new and existing customers and in all of our regions. And that is also something that is very much reflected in the numbers for the last quarter. Order intake increased by 8% in the fourth quarter and 17% for the full year. And I want to highlight that these numbers include the effects from the stronger Swedish krona, which, as you know, has affected us a lot during the year. And that goes both in terms of growth and on the bottom line. Order backlog end of December amounted to 1,528,000,000 and the book-to-be ratio was 0.9 both in the quarter and for the full year. Moving on to net sales, we finished the year beating the last sales record by far. Net sales of 375 million means a growth of 16%. The organic growth was 28%, showing that the underlying business really performs well. Licensed sales growth was 15% in quarter four, and support sales grew with 6% year on year. When we take out the currency effects from the support sales numbers, the growth was 16%. The high net sales drove EBIT to 92 million a quarter and strengthened the margin to 24%, compared to 23% for the same period last year. Currency losses from the revaluation of working capital affected EBIT with just about 10 million, and adjusted for that, the EBIT margin would have been 27%. Next slide is the rolling 12 development of net sales and EBIT and the perspective that we believe gives a better description of research business performance. For the full year 2025, net sales increased 13% to 1,344,000,000. The organic growth was 19%, and with an EBIT of 292 million, we ended the full year 2025 with a margin of 22%. That is equal to last year, but also burdened by 37 million in currency losses. Adjusted for those and an additional 23 million that we treat as non-recurring costs, the margin would have been 26%. Moving to the next slide showing the revenue split and where I focus on the revenue from support, we saw a growth of 11% in our support revenue for the full year 2025. With the steady growth we have in our support revenue over time, we increased the robustness in the business from recurring revenue. And for the total year 2025, the portion of recurring revenue in relation to total net sales was 39%. Cash flow in quarter four, as you can see here on the next slide, improved significantly and amounted to 91 million. And that includes positive effects from a lower working capital. We will continue to put focus on having a good cash flow in 2026. However, I want to point out that I also, or what I also said in quarter three, that the cash flow can fluctuate also going forward. We always seek to work with standard payments or standard payment terms in our customer agreements, but we also have situations where the gap between sales and payment is longer. It can be tenders or framework agreements or related to certain markets. Sales that comes with a good profit, but where we have to accept later invoicing. We want to have a position where we sometimes, for strategic reasons and in relation to important customers, can choose to accept profitable sales over short payment terms. And with a cash balance end of 2025 amounting to $407 million and no loans, we have a solid financial position. The next slide shows the contract assets, that is our customer receivables, and also our contract liabilities. That is the balance sheet items that shows how much payments we have received from our customers in advance. We have during 2025 moved away from a position where our contract assets were lower than the contract liabilities. And that is to a large extent dependent on that we have delivered on prepaid sales in our backlog. But I want to point out that a net position of 118 million is still a good position. But of course, this doesn't take away our intention to lower this number and to improve the working capital where we can during 2026. And with this, I hand over back to you, Johan.

Thank you very much, Nina. So I will just briefly mention the dividend proposals. We are pleased to announce that the board proposes a dividend of four Swedish crowns per share for 2025, which is up from three crowns per share. The dividend will be decided at the annual general meeting on May the 7th. The board has also revised research dividend policy effective from 2026. The goal is to distribute 50% of profit after tax annually, taking into account the company's capital needs, investment opportunities and overall financial position. And I would like to devote some time to AI and how it affects research. It's very important to note that AI is something very positive for research and it's definitely not a threat against our products. There has been some belief in the community in general for software companies that AI can create and replace ordinary system development. That's probably true for simpler applications and with thin functionality and not so much data. With our large and complex systems, it's not doable for AI today. AI can only produce smaller snippets of code with high quality. Also, in our field, we need deep domain knowledge. We also need to consider patient safety as well as cyber security and we are liable for that and we have to take responsibility for the code. AI could never make sure or promise that there is no maltreatment of patients for example. So as a company we need to understand the code and make sure that it doesn't harm any patients as we treat millions of cancer patients and we cannot make a mistake one single time. There is also huge data requirements in our field. We need clinical data, images, plans, contours, etc. We need to perform measurements for machine modeling and quality assurance. Then we have the medical device regulations such as FDA. We as a company have to promise and document that our system performs according to the requirements and that it is a safe application. And AI doesn't take any responsibility in that regard. And it's also, we are existing in an ecosystem with many, many partnerships with machine vendors and our installed base of about 1,200 clinics. And in order to develop these platforms that we develop, we need to do that in partnership with all of these stakeholders. So AI for research is a very useful thing. We have a large machine learning department at research where we leverage AI for our products. For example, we have a functionality in RayStation called deep learning segmentation where we based on images such as CT images and MR images can automatically segment the organs in the patient as you see in that image to the right. So those are about 200 structures in the patient that has been automatically segmented with deep learning segmentation, and it takes about one minute to do that, which would take many hours to do in a manual setting, or in a manual manner. And this is used clinically throughout the world, and only 2025, 270,000 patients were segmented using this particular module. This leads to significant time savings and it also increases the segmentation quality and you can achieve better consistency over different users and over different institutions. The second product that we have in RayStation is deep learning planning. So here we automate the very time consuming task of treatment plan generation. 7,000 clinical treatment plans have been generated by our customers so far, but this is increasing rapidly now as more and more customers get their hands on this technology. This increases plan quality and, again, consistency and saves a lot of time. It also opens up for multiple treatment plan generation per patient so that we can explore a larger solution space. One good example is a customer in Belgium, Iridium, that have now automated almost all of their prostate patient planning using the AI capability in RayStation. So what they have seen is that the deep learning planning models outperform manual planning by a human being, achieving superior quality and consistency. And you can see some of the time savings that they achieve. So on the patient modeling side, they save 44% time. And on the plan generation side, they save 47%. We also use AI to help our developers write the code faster. So AI can then, for example, Microsoft Copilot, can help our developers to find bugs, can explain complex code and patterns, write tests, and also help with documentation. But it's important that developers stay in control. They always review and modify the AI output. the code that's generated by AI is not always is not very tidy or beautiful so we have to the developers have to stay on top of that okay and now the final section of the presentation is a quick summary and outlook so we saw that we had record high net sales despite macro uncertainty and very heavy currency headwind. In spite of that, we could show solid profitability and also improve cash flow, which we are very happy about. There is still a strong demand for research solution and increasing demand, I would say, for research solutions across all the regions. And we are confident about our EBIT margin target of at least 25% in 2026. So with that, I will open up the Q&A session, and I believe Carolina will manage the questions.

Thank you, Johan. Yes. We will start the Q&A session with live questions, but before we do that, I would like to remind you that you can also post written questions in the Q&A chat. So let's start with the first question that comes from Christoffer Liljeberg at DMV Carnegie. Christoffer, please go ahead and unmute yourself.

Yeah, there, sorry. I have quite a number of questions. Maybe I'll start with three and then come back. So first... Can I ask you to ask one question at a time? Okay. I'll do that. You're the boss. Maybe then I would like to ask about the support revenues, if there are any one-offs here helping that in Q4, or if that's a good starting point for 2026?

Yeah, that's a question for me then. Yeah, we have some one-offs. It is not very much. But, I mean, it is a little bit tricky, I think, to talk about one-offs in our support revenue because we always have a little portion of that. We have situations where... Our customer contracts are, I mean, there is a delay in timing when they are renewed. And it might be that we, because of that, have revenue for, I mean, more than three months in one quarter. So... It's a little bit too hard to say, Kristoffer, give a straight answer to that. But I would say that you can use this as the base going forward.

Great. Thank you.

Please go ahead, Kristoffer.

Okay, okay.

No, no, it's one question at a time, that's all.

Okay, okay, that's helpful, thank you. Yeah, my second question, the news that you sent out a couple of weeks ago about Royal Morse and doing online adaptive on... electa machine was this without ray care and if so how are they able to do that i don't know if that's if it's possible to just give a quick quick answer on that yeah um there was some versa hd like the machine so far only ray station without raycare but that means

um the workflow is somewhat clunky takes more time um and uh but it is doable uh to do it and that's the the important message here they will um implement raycare going forward and then the workflow will be smoother and quicker of course it's um it's more They have also a Radixact machine, so we'll be quicker on that machine, given that we have interoperability between Raycar and the Radixact. And it will be also smoother on a true beam, a variant true beam, since Raycar is fully integrated. But the point here is that even without this strong integration, you can do it, but it's not as quick.

Okay, that's helpful. Thank you. And my third question, if you could comment on the Pinnacle conversion in Q4 and outlook for that here in 2026.

Yeah, we will of course focus, this is the last year that Pinnacle is around, so there will be a strong focus during 2026. In Q4 it was actually surprisingly low. It has been a quite high percentage of license sales in previous quarters. In Q4, it was actually the license revenues from pinnacle conversion was only 11%. So that shows that we can, because I think that has been discussed, and there has been a lot of questions about whether we are able to convert other clinics than pinnacle clinics, but that shows you that that's very possible.

Thank you very much. I get back to the queue and let someone else ask questions. Thank you.

Thank you, Kristoffer. The next question comes from Mattias Wadsten at SEB. Mattias, please go ahead and mute yourself.

Good morning. Can you hear me?

Yes, we hear you. Love and care.

Good. I will always take them one by one. So... You shared a license share of Pinnacle here in Q4, which was a low number. Could you share that for the full year? And also, that leads me to believe then that the conversion, the Varian conversion and Lecter conversion must have been very strong to end the year. So just, yeah. If the momentum has switched gears there, and what's driving that? That's the first one.

Yeah, okay. So... First, for the full year number, I think it was 43% licensed revenues from thin acne conversion. Yeah, it's just that we have been able to convert other types of clinics. For example, this large UPenn University of Pennsylvania order in Q4, which was, I believe, 57 million Swedish crowns, around that number. Yes.

I'm not hearing revenues in order.

What was it? 53, then, in revenues?

A bit above 40.

Aha. Okay. And that was an eclipse conversion, so that, of course, affected that mix for the Q4. So this will... vary from quarter to quarter. It's very hard to predict. We will have, since we have a time-limited opportunity now for pinnacle conversion, there will be strong focus for that in 2026.

Good. And then I have a follow-up on Christopher's question on the online adaptive radiotherapy that you can perform on Electa, Linux, and TrueBeam with RayStation. But do I still read you correctly that in order for a clinic to seamlessly sort of perform online adaptive, you would still need RayCare in the future, or how should I interpret that?

That's correct. To have a broad clinical use for this, RayCare is needed. So you have understood that correctly.

Okay, good. And do you expect it to be frequently used among those clinics that have RayCare, you know, for maybe 2026 and years from?

I think the main drivers for RayCare are going forward now that we have a very good combination of equipment with RayStation RayCare and various TrueBeam, where you can make extremely effective online adaptive treatments. So we see a lot of it... all over the world for this combination.

Okay, thank you very much. I will limit myself to one more question. So in terms of the new orders, the University of Pennsylvania, if that was recorded as sales in Q4, and then maybe the same question for Greater Poland Cancer Center as well. That's my last one. Thank you very much.

Yeah, it was. A big portion of the European order was recorded as sale, as Johan also just said.

Perfect. Thank you. Thank you.

Thank you, Mattias. We have a question from Oskar Bergman at Red Eye. Oskar, please go ahead and unmute yourself.

Thank you. Yeah, just wondering if you can give an update on roughly how many pinnacle centers are left to convert, and also then, were there fewer conversions in absolute terms from the clinics in Q4, or have you sort of lost in the market share now on conversion?

Okay, we don't know the number of pinnacle clinics. It's in flux right now, so it's very hard to know the number of remaining pinnacle clinics. In some countries, there are almost none, like in the UK and Japan. They have been basically all converted. In Germany, there are quite a few remaining in the United States and China, but it's a couple of hundred. I can't give you more detail, but it's still a big opportunity out there. And no, we have not lost market share in terms of the Pinnacle conversion. It's rather that other conversions have been, we look at percentages here. So in absolute numbers, we haven't, we're still very successful in converting Pinnacle clinics to RayStation.

Okay, thank you. I always ask about RayCare, and I have to ask about it also this time. I'm just wondering how many new RayCare centers were signed in Q4, and perhaps also if you can elaborate on what remains the largest obstacle for increasing RayCare clinics.

There are no real obstacles. We had, I believe, four RayCare orders in 2025 in total. Of course, that's not where we want to be, but we believe that this will ramp up during 2026. Okay, one obstacle is the online adaptive capability, which needs two new versions of RayStation RayCare. It requires FDA approval, and that takes the time it takes. It's not something we can It became affected to some extent, but it was also in the hands of FDA. So that's needed, but in Europe, the online adaptive treatments on this platform will start during spring. So we see the first, and that's also... a good message for the U.S. market, because then they see it's just a matter of time before they can get their hands on this functionality as well. So regarding ramp-up of Raycare, it only takes time. There is a lot of interest for Raycare now. There are no particular obstacles in place, so we are quite confident that we will see Let's say over the next two or three years, a good ramp-up of regular sales.

I think in the Q3 report you mentioned that you opened up some modalities for a customer base for a six-month trial period. I'm just wondering if we can get an update on how that has progressed so far.

It's still limited to a couple of countries, and it's progressing well. So they are very happy that they can try out new functionality. I think the limiting factor there is the time they have at their disposal. They're running very busy clinics, and it's hard to spend time on just exploring new functionality. But otherwise, it has been very well received in the countries where we have opened up so far. Okay, thanks.

And just a final question. I know you're not supposed to give your view on the share price, of course, but at these share price levels, why are you focusing on dividends rather than stock repurchases?

That's a good question. I think buying back shares is an interesting option that we will look into deeper. So we are looking into that, for sure. All right. Thank you very much. Thank you.

Thank you, Oskar. We will now take a question from Arjan Nordarmer. Please, Arjan, go ahead and mute yourself.

Hello. Can you hear me? Yes. Yes. I have a question about the order backlog. So we have seen it steadily decrease over the past few years, and right now is under 1.1 times for the sales for this year. Why is it decreasing so much? Is this like a problem for revenue growth going forward, or is there something else going on here?

The main reason why it has shrunk lately is the dollar effect or the currency effect. So, no, we don't really see it as a problem.

Okay, so you don't think that is limiting growth, like over the past few years? No. Okay, thank you. That was my question. Thank you.

Thank you. We will now move back to Christoffer Lilleberg at BMW Carnegie. I guess you have a follow-up question.

Yes. Yeah, a few more. First, just a clarification, the 11 of 23% you mentioned for Pinnacle conversion part of total licenses, is that total licenses or licenses to new customers?

Out of total licenses.

Yeah, great. Thank you. Then a question on the cost, particularly administration costs seem to have remained high here in Q4. Sequentially given the, I guess, third quarter you should have had the extraordinary cost, much of that in that line or?

Sorry, Kristoffer.

If I look at the administration costs, they remain at a quite high level. They're actually higher in Q4 than in third quarter and second quarter when I guess you had costs for the employee conference. Or was that in another cost line?

No, it's included in the administration costs.

Okay, but did you have such costs this quarter as well, or why does administration costs remain so high?

No, we didn't have those costs in this quarter, and yeah, it's a good question. I must come back to that one. I haven't looked at it from that perspective.

Okay. And then maybe one, I don't know if you want to say, you sound pretty positive in the CEO word in the report. So when it comes to the sales outlook for 2026, do you expect a similar positive trend here or anything that could change that?

Thank you. No, to achieve the 25% EBIT margin, or at least 25% EBIT margin, that relies heavily on sales growth. So we are positive in that regard.

Great. Thank you.

Thank you.

Now we have a question from Mats Andersson. Mats, please go ahead and unmute yourself.

Hi, I have a question about the Ortega order. In Q3, you said that the first center will have your income in Q4. So my first question is, how much is your income in Q4? And when will the next delivery to the next center go? I didn't hear which center... Was it Ortega you were talking about? Yes, Ortega order.

Got it, got it. That become, there hasn't been any revenue from that during 2025. But our estimate is that there will be revenues from at least two centers during 2026 from the Ortega order. Then that will be delivered and booked as revenue.

Okay, thank you. Thank you.

Moving on to the next question that comes from Carlos Moreno. Can you please unmute yourself?

Hello. You're obviously very near your kind of previously set medium-term margin targets. And you mentioned in Q3 that, you know, you might revise those targets, give new long-term guidance. Do you still expect to do that at some point during 2026? Yeah.

During 2026, we will communicate a new plan. let's say, three-year margin target and possibly some other financial target. But you can expect that will be communicated.

And is that with, like, the half year or the first quarter or some time?

I don't know for sure. It involves the board. The board has to make a decision. So I can't say by myself. But... That's not a problem. We want to communicate a new, let's say, medium-term target.

Sorry, I interrupted you.

I'm sorry, say again? Sorry, I interrupted you. I apologize. Sorry. No problem. Go ahead. No, that was it. That's good. So we're going to get some new targets sometime during the year. By the sound of it, we're going to get margin and maybe sales. There's going to be some sort of more than margin medium term target. OK. And I just want to add what the another person said. I mean, if your shares are just being pushed down because they're in some basket, I appreciate the dividend is fantastic. But it just seems to me you have to be on the other side of that. AI selling and it just seems to me a buyback is very, you know, there'll come a point where spending your cash on buying your shares is a very sensible investment, right? And to me it just seems like an extremely good idea. But anyway, I just wanted to look at that.

I note your comment and I think you're probably right. Thank you.

Thank you. And we have a follow-up question from Arjan Nordermere. Arjan, please go ahead.

That wasn't excellent, sorry.

Okay. Oh, we have a follow-up question from Mattias Badsten. Mattias, go ahead.

Thank you. I just thought that you could help disclose some outlook on timing of approvals, release of modes to expand the use of the software products you have to further cancer therapy areas.

My name for that is so if take liver ablation for example that can be used in Europe as of now. There we are waiting for 5-10K clearance in the US. Chemotherapy will be cleared sometime during 2027. And surgery will be, yeah, that's even further into the future, so I can't say that. So liberation will be first, chemotherapy after that, and then surgery is coming after that.

That's very clear. Thank you.

Thank you.

Thank you all for your questions. With that, we will conclude today's presentation. A recording will be available shortly on our investor website, and if you have any additional questions, you are very welcome to reach out to us. We appreciate your participation today, and we look forward to connecting with you again on April 29th when we present our Q1 results. We wish you a pleasant rest of your day. Thank you. Thank you.

Thank you.