Raysearch: Laser Focused Oncology software provider RaySearch ($RAYB.st) is down almost 50%, caught in the middle of the selloffs in Software and in Swedish healthcare stocks, and with Iran war volatility on top.
By Myles Kuah
11 min. readView original Oncology software provider RaySearch ($RAYB.st) is down almost 50%, caught in the middle of the selloffs in Software and in Swedish healthcare stocks, and with Iran war volatility on top. I believe that this has provided us with one of the most attractive opportunities I’ve ever covered here and I want to go into some depth about their product, and more importantly their interesting competitive position within their markets. While the market is worried about AI disruption, I believe their product has significant barriers to entry that protect it from the issues most software companies are facing, while holding a near-monopoly position in one of the fastest growing corners of oncology.
The Business RaySearch Laboratories is a Swedish medical technology company founded in 2000 as a spin-off from the Karolinska Institute in Stockholm by current CEO Dr Johan Löf. Put simply, RaySearch builds complex mathematical algorithms for radiation therapy. These algorithms model patients anatomy and calculate the exact angle and strength of radiation beams to treat cancer. Löf has a PHD in Medical Radiation Physics, and founded the company with the premise that the people best positioned to write these algorithms, as opposed to the hardware companies which dominated the space. These algorithms can be the difference between life and death and have huge reputational and compliance barriers to entry that limit competition to only a few companies.
Raysearch’s revenue comes entirely from software sales and support revenues. In 2025 48% of their revenue came from one off licence sales. However, these licence sales work as an installed base, with 40% of their revenue coming from recurring support agreements which include things like software version updates (which includes essential regulatory continuances), customer support and new model (hardware) support. This recurring revenue is incredibly high margin and comes in at essentially 0 variable cost. The other 12% of revenues consists of lower quality hardware and training revenues. This is a fantastic business model as RaySearch continues to build up a consistent, reliable source of recurring income that makes them more resilient against market downturns or weak years.
At this point I need to make sure to distinguish between the two core types of radiation therapy that RaySearch deals with: photon and particles. Photon therapy uses linear accelerators (LINACs) with high energy x-rays to treat the cancer. Photon therapy is less precise and has more side effects due to the photon’s lack of mass, which makes it move through the body and affect surrounding tissue. In spite of this it is the global standard for radiation therapy and accounts for around 85% of the radiation therapy market due to being significantly much cheaper (25x) historically, though this is changing. In contrast to this, particle therapy uses particles like protons (by far the most popular), carbon and helium, which due to their mass are able to perform a targeted energy release at a more precise location. While particle therapy (most notably proton therapy) is much less popular, the last decade has seen the development of smaller, cheaper compact systems that bring the price from around $150m down to $25m which has seen strong persistent growth in new proton therapy centers.
Competitive landscape This distinction is important to note as RaySearch holds a very different competitive position in these two markets. Within the broader photon market there are two core competitors in Varian’s ‘Eclipse’ system (owned by engineering giant Siemens) and Elekta’s ‘Monaco’ system. The literature here is inconclusive as to whether RayStation is more effective than Eclipse of Monaco from a treatment perspective. Both of these companies sell the systems used in photon therapy and bundle the software in with them, making it a naturally challenging, but not impossible competitive undertaking for RaySearch to compete. In contrast, within particle therapy RaySearch holds an almost monopolistic 80% of the Proton therapy market. The core difference here is that while Varian and Elekta hold 75% of the market in photons, Varian holds a much smaller percentage of the market in Proton therapy hardware, and other competitors like IBA, Hitachi and Mevion don’t have their own software and so use RayStation almost by default. When looking at their prospects going forward we need to look at these two markets separately as they have very different opportunities. I want to note here that it is incredibly difficult to envision these markets being meaningfully disrupted. These software offerings are built off decades of mathematical optimisation with both data and expertise that would be incredibly difficult to replicate. Even if a competitor built software as effective, they would have to pass regulatory hurdles along with beating the reputational inertia and switching costs to compete for what is a fairly small and niche market. The incentives just aren’t there for meaningful competition even with the lowering barrier to entry into software with AI, because it isn’t the software layer that gives RaySearch the edge.
The photon therapy market is a genuinely competitive and difficult to breach one for RaySearch. As stated earlier, Varian and Elekta dominate the space with around 90% market share in hardware and 75% market share in software, mainly due to packaging their software with their hardware. These types of software are very sticky, and it’s rare for a clinic to change software as that would involve their operators being forced to retrain on new software and more importantly replanning every existing patient’s. So how does RaySearch win market share? Due to having no dedicated hardware offering, RaySearch has by far the best software offering for multi vendor clinics. Both Eclipse and Monaco are fantastic with their native hardware, but are notably worse when used with machines from other providers they weren’t built for. For any clinic that uses machines from different hardware providers, or wants to be flexible going forward RayStation is by far the best offering. Additionally, for the small segment of the market not using Varian or Elektra, RaySearch is almost the default choice. A current tailwind is the discontinuation of competing software Pinnacle by Philips in December 2026. There are hundreds of clinics worldwide using Philips photon therapy software that are having to move to new software, and many of them are choosing Raystation with Pinnacle conversions accounting for 23% of total licence sales in 2025. Altogether the photon therapy market is expected to grow at around 7% over the next few years which makes it a growing market. For photon therapy, RaySearch doesn’t need to take a huge amount of market share to do well, but just to continue to succeed in their existing niche.
Additionally to this, a structural growth opportunity in photons is emerging in online adaptive radiotherapy (OART). Conventional radiotherapy uses a fixed treatment plan with large safety margins to account for daily anatomical changes like tumours shifting, organs moving and patients losing weight. OART eliminates this by replanning in real time while the patient is on the table, delivering more precise dose to the tumour and less to surrounding healthy tissue. Until 2025, OART required purpose-built machines costing $8-15M, limiting it to a handful of elite centres. RayStation is now the first and only commercially validated software to enable OART on standard LINACs, meaning any of the 8,500 existing LINAC centres can access it via a software upgrade. RaySearch is 2-3 years ahead of any competitor here, with a captive upsell opportunity across their entire installed base and a natural pull-through into RayCare adoption.
The proton therapy market is where the meat of this thesis is. As stated earlier RaySearch holds 80% and growing market share within this sector, making it essentially a monopoly. The growth narrative behind this sector is real, with growth estimates between 10-12% for the next 5 years as clinics continue to take advantage of the lowering costs of proton therapy systems. What’s notable here is that while RaySearch currently has more photon customers, their proton sales are much higher value per customer. As mentioned earlier, the cost of a proton therapy system is around 5x the cost of a photon therapy system which means they can comparably scale up the price of the software, especially as they are the only serious option available. While they don’t break out specific values for proton vs photon sales, based on the number of clinics it’s fair to estimate that around 20-35% of their revenue comes from proton software sales, however this percentage will continue to grow as the growth in proton clinics continues to outpace the growth in traditional photon clinics.
One thing to note here is the optionality in broader software offerings RayCare and RayIntelligence. RayCare is an oncology information system, which you can essentially think of as an ERP like ServiceNow but hyperoptimised for workflow within the oncology clinic. Meanwhile RayIntelligence is a cloud-based analytics platform for analysing data from RayStation and RayCare. The core offering here is an ecosystem, ensuring that all of a clinic’s software works together seamlessly with specialised offerings. These offerings currently aren’t huge revenue contributors, but are very easy, high margin upsells to new RayIntelligence customers. In their most recent earnings call, management stated that in the next 2-3 years they expect to see “a good ramp-up” of RayCare and RayIntelligence sales. I’m not really factoring a big expansion of these offerings into my valuation model, but if there were that would provide meaningful upside.
Management I don’t believe management here leaves anything to complain about. As mentioned earlier the company is founder-led by Dr Johan Löf who has done a fantastic job of building the company, up 41,000% since listing in 2003. Löf is well incentivised to perform well, owning 10% of the company and over 70x his annual compensation in stock. Management is stable, with most members having been around for a reasonable amount of time. Management and board members own a significant amount of stock with 7% of the company held by other insiders. There’s even been a recent insider purchase with a member of the board buying $280k USD worth of stock around the current price.
Valuation While this is a very attractive business to me, I am still a value investor to my core so valuation matters. This is where things get interesting. At a 7700 mSek market cap and 244 mSek underlying earnings (adjusted for receivables currency movements and fully expensing R&D) RaySearch is currently trading at a reasonable 31x earnings. However, when we model out the numbers this gets far more interesting as this is still a company reasonably early in its development. The important thing here is that at 90% gross margins and 22% operating margins made up mostly of fixed costs the company has a huge amount of room for margin expansion. In the last 3 years as revenues have increased from 843m in 2022 to 1350m this year, operating expenses have only increased from 718m to 950 seeing operating margins expand from 4% to 24%. With limited variable costs, it’s reasonable to expect operating margins to continue to trend upwards and CEO Löf has set 25% as a floor for their operating margin for 2026, however I believe there’s a reasonable chance it’s a bit higher. The company has been putting up low double digit revenue growth for a while (19% currency adjusted last year) and I think that’s a reasonable expectation going forward with the exception of 2027 which will have the Pinnacle growth cliff. So to run some potential numbers, if we were to see 17% revenue growth (which would be low historically) next year at a 26% operating margin we’d end up with around 323m of earnings, putting us at a forward pe ratio of 24. I’m not saying that those are the numbers that they are going to put up, but that’s a reasonable estimate that puts us at a valuation that is too low for a business of this caliber. I’ve done some loose financial modelling (they don’t give a lot of granular details so it’s very loose), and the base case for my different outcomes seems to fall around 50% upside to the current valuation without super demanding assumptions.
Looking at the other financial statements, the company continues to be spotless. RaySearch has no debt, and 400 mSek net cash. Their cash conversion is reasonable and while they do capitalise their R&D expenses, the D&A has caught up and tracks their capex closely with only around a $15m difference last year. Regarding capital allocation the company has a dividend policy of 50% of profit after tax annually. They have not historically engaged in significant m&a so aren’t likely to blow their profits on a stupid acquisition, but also I am interested to see if they have any more detailed capital allocation plans going forward now that they are consistently profitable as they are beginning to build up cash on the balance sheet.
Risks The core question I like to ask when doing these writeups is “why does this opportunity exist?” I think in this case this question is pretty straightforward to answer. RaySearch is down almost 50%, caught in the middle of two huge struggling sectors in Swedish medtech (which has been significantly underperforming the broader Swedish stock market) and of course the large software selloff we’ve seen globally due to AI fears. The Swedish medtech selloff is a purely valuation based one, meanwhile I believe that RaySearch should be immune to long term disruption from AI due to many of the reasons mentioned earlier, so I see them as a diamond that has been unfairly dragged down to unreasonable levels due to sector wide indiscriminate selling.
There are two other factors impacting the stock that I see as inconsequential long term. The first of these is currency impacts. While organic revenue growth was 19% in 2025, when accounted for currency movements it came in at 12% due to the appreciation of the SEK against the USD, plus a $30m currency related decline in receivables. This has resulted in earnings and revenue growth being understated relative to the underlying business performance, but isn’t hugely consequential for international investors. The other factor is revenue lumpiness. As a large proportion of their revenue is licence revenue tied to large projects their quarterly and even annual revenue has been and will continue to be lumpy. The main thing to focus on here is the long term performance of the business.
As far as the actual risks there are a few. The most pressing is in photon therapy. While they have consolidated market share well, their market position as the multi-vendor software of choice could be under threat with Elekta specifically targeting that area of the market with their Elekta ONE planning software released in 2024. While RaySearch has held them off well, this is probably where they’re most vulnerable. Another notable risk is that growth in proton therapy clinics isn’t as fast as expected. This could be due to some sort of broader global growth slowdown, or maybe just estimates are too aggressive. In this scenario, services revenue should provide some cushion but returns would still likely disappoint. While any of these downside scenarios would hurt the stock, the combination of growing market, recurring revenue, and undemanding valuation I believe limits the downside and likelihood of permanent capital impairment.
Conclusion To be frank, I believe that RaySearch may be the most interesting idea I’ve ever written about on here. In RaySearch you have a genuine software monopoly in the growing proton therapy market combined with a high quality, competitive offering in photon therapy succeeding within a niche. I think our downside is reasonably protected by switching costs, regulatory and reputational moats, recurring revenue and a reasonable valuation. The current selloff in software and Swedish medtech has provided us with an opportunity to buy a low double digits organic grower at 24x forward earnings, a price I believe to be too cheap considering the many factors discussed above. For full disclosure, I have made RaySearch one of my largest positions with an average price around 220 mSek.