[RAY B] RaySearch Laboratories: The Software Nobody Wants to Need

Inside RaySearch's counter-intuitive business model of selling what competitors give away free

Origins

In 1986, as Sweden's universal healthcare system earned international acclaim for its comprehensive cancer care, seventeen-year-old Johan Löf sat beside his mother as she died from the very disease it was designed to defeat. The irony was personal before it became instructive. Even the world's most advanced medical infrastructure couldn't solve cancer's fundamental challenge: precision.

Radiation therapy could destroy tumours, yes, but it butchered healthy tissue in the process. For a mathematically gifted teenager drawn to physics, this wasn't inevitable tragedy. It was a solvable problem.

Löf's subsequent journey through engineering physics at KTH and his PhD work at the prestigious Karolinska Institutet positioned him at the epicentre of a quiet revolution. Under Professor Anders Brahme's supervision in the late 1990s, he developed algorithms that could optimise radiation delivery with unprecedented precision, mathematical models that customised treatment plans for individual patients. Desktop computing had finally reached the threshold for complex optimisation. Medical device companies were beginning to grasp that software, not just hardware, would drive the next generation of breakthroughs.

When Inverse Problems published Löf's research, it validated both scientific rigour and commercial potential. The transition from academic breakthrough to business opportunity needed only a catalyst, which arrived via Sweden's Venture Cup competition. Löf participated reluctantly. He had, by his own admission, "no thoughts on starting up a business."

Perfect.

The founding of RaySearch Laboratories in 2000 exemplified the Swedish approach to innovation: patient, methodical, grounded in genuine technological advantage rather than market theatrics. With just SEK 6 million from Affärsstrategerna, modest even by Swedish standards, Löf made a decision that would define everything: maintain independence at all costs.

This apparent constraint became liberating. While competitors burned cash fighting each other, RaySearch quietly made itself indispensable to all of them. The partnership strategy proved prescient when Philips signed on just two months after founding, integrating RaySearch's algorithms into their treatment planning systems. By 2001, 131 systems had reached clinics worldwide. A small Stockholm company achieved global reach while building the foundation for future independence.

But Löf recognised that ultimate success required controlling the entire experience. As partnerships expanded throughout the early 2000s, Varian, Nucletron, TomoTherapy, the fundamental tension emerged: remain the brilliant partner, or become the direct competitor?

The 2008 decision to develop RayStation marked the pivot. Launching during the global financial crisis, when sensible companies were retrenching, demonstrated the confidence that comes from deep technical moats and patient capital allocation. When RayStation received FDA clearance in 2010 and Massachusetts General Hospital became the first US customer, it validated more than technology, it vindicated the strategic vision of a Swedish academic who had transformed personal loss into a tool for saving lives.

Today, with RaySearch's software used in over 2,600 clinics across 65+ countries, the company represents something increasingly rare: authentic mission meeting technical excellence, sustained by the kind of strategic patience that turns reluctant entrepreneurs into quiet revolutionaries.

How do they make money

RaySearch's customers pay millions for software they desperately hope will become invisible.

The company makes treatment planning software called RayStation that helps medical physicists design radiation therapy treatments for cancer patients. Think of it as sophisticated 3D modelling software that calculates exactly where to aim radiation beams, how intense to make them, and how to shape them around tumors while avoiding healthy organs.

They also sell RayCare, which manages patient information and clinical workflows throughout cancer treatment. Hospitals buy these systems, along with ongoing support contracts, training services, and specialised computing hardware.

The software does the complex physics calculations that determine whether radiation therapy succeeds or fails, which is why customers treat purchasing decisions with life-or-death seriousness.

But what RaySearch actually sells isn't software, it's the ability for cancer treatment professionals to sleep at night.

When the algorithms work perfectly, medical physicists forget the software exists, it simply calculates where to aim radiation beams that destroy cancer while sparing healthy tissue. When it fails… people die. This invisibility paradox drives everything: hospitals will pay premium prices for technology that disappears into critical workflows, while medical physicists stake their professional licenses on calculations they cannot verify independently. In 2024, this professional assurance generated SEK 1,192 million across four carefully orchestrated revenue streams.

One needs to understand the scale of institutional commitment. That SEK 77 million Heyou Hospital order, enough to buy a small Swedish castle, represents algorithms that will calculate radiation doses for Chinese cancer patients who will never know this software exists.

Individual transactions typically range from hundreds of thousands to tens of millions of SEK, but, interestingly, hospitals representing barely 13% of the world's 8,500 radiation therapy centers are already paying RaySearch SEK 400-600K annually per clinic in support fees.

License sales (48.3% of revenue) appear to drive growth, but support contracts (39.7%) create the ongoing relationship. Hardware sales (9.2%) and training services (2.9%) complete what looks like conventional software-as-a-service. It's more sophisticated than that.

What customers call "support contracts" is actually medical advancement subscription which serves as technical assistance. Here's the mechanism: cancer treatment evolves continuously through new techniques like adaptive replanning, proton therapy protocols, and AI integration. Without current support contracts, centers cannot access these innovations, cannot integrate new equipment from Varian or Hitachi, cannot maintain regulatory compliance. You're not buying software maintenance, you're subscribing to the cutting edge of cancer treatment.

Once RaySearch software integrates into clinical workflows, switching becomes extraordinarily difficult. Training medical physicists requires 6-12 months while they remain responsible for active patients. Institutional protocols, quality assurance procedures, and regulatory validations all build around specific software interfaces. Equipment integrations with CT scanners and linear accelerators require custom configuration. Most significantly: existing patients have treatment plans stored in current systems, preventing mid-treatment transitions.

When Memorial Sloan Kettering's medical physicist reviews treatment plans, she's protecting her license, her career, and patient safety. RaySearch has 1,100 customers facing similar stakes, each committed by the mission-critical nature of their dependence.

Numbers

RaySearch's customers pay SEK 473 million annually for the privilege of being trapped, but it's a golden cage they willingly enter.

Revenue composition in 2024 reveals the elegant machinery of medical software dependency: license sales contributed SEK 575 million (48.3%), while support contracts delivered the real prize, a recurring stream that grew 17.3% while maintaining its substantial 39.7% share. What is not obvious is that conventional software revenue models pale in comparison to what RaySearch has achieved. Support contracts represent continued access to life-saving algorithms that become so deeply embedded in clinical workflows that switching becomes virtually impossible, creating dependency through excellence rather than coercion.

Three years ago, RaySearch was bleeding money at -8.3% margins. Today, it prints 21.9% profits on incremental sales. The software business model's mathematics finally kicked in, transforming what appeared to be gradual improvement into dramatic profit expansion.

Operating profit reached SEK 260.5 million, and first-quarter 2025 results showed an even more impressive 28.6% operating margin when adjusted for currency effects. By any measure, this represents the extreme operating leverage inherent in software businesses once they achieve scale, where incremental revenue drops almost entirely to profit since the marginal cost of additional licenses approaches zero.

RaySearch performs accounting alchemy, transforming SEK 197 million of expenses into 'assets' with a few keystrokes. The company invested SEK 279 million in R&D, 23.4% of revenue, but capitalises 71% of development costs rather than expensing them immediately. They're either strategic visionaries or aggressive accountants, and the difference matters for investors.

The reported 21.9% operating margin masks an economic reality where true cash margins approximate 5-6% when all R&D flows through the income statement. In other words, RaySearch is borrowing profitability from the future, betting that today's algorithmic innovations will generate returns three to five years hence.

Hospitals essentially lend RaySearch money to build software they'll buy later, and pay handsomely for the privilege. Operating cash flow of SEK 485 million represents a 41% conversion rate from revenue, demonstrating the cash-generative power of this business model. The apparent contradiction, cash flow nearly doubling operating profit, reveals customer prepayments for multi-year support contracts creating a working capital advantage where RaySearch borrows from customers to fund operations. This metric matters because it strips away accounting conventions to reveal actual liquidity generation.

SEK 466 million sits trapped in healthcare bureaucracy, 142 days of revenue tangled in procurement committees, regulatory approvals, and budget cycles that move like continental drift. While the collection period initially appears concerning, healthcare institutions operate with extended payment cycles due to complex procurement processes, making this standard rather than problematic. Yet on closer scrutiny, this represents nearly 40% of annual revenue tied up in customer promises to pay, creating both credit risk and working capital demands that could strain cash flow if economic conditions deteriorate.

RaySearch has built a subscription business that doesn't know it's a subscription business. The order backlog totalled SEK 1.81 billion at year-end, but what matters is that support commitments represent 72% of total backlog, meaning SEK 1.3 billion in contracted future revenue from existing customers who've already committed to multi-year relationships. With slight exaggeration, this contracted revenue base functions as voluntary dependency, customers paying recurring fees for mission-critical software that cannot be easily replaced without significant disruption to patient care.

In just twelve months, net debt collapsed from SEK 186 million to SEK 9 million, transforming the balance sheet from leveraged growth story to cash-rich operation. Combined with SEK 463 million in cash, this financial position eliminates refinancing risk while providing strategic flexibility. The result came from strong cash generation rather than equity issuance, indicating organic business improvement rather than financial engineering, a clear sign of business model maturation.

RaySearch trades naked in currency markets, deliberately exposing SEK 57 million of annual profit to foreign exchange movements. Each 10% movement in USD/SEK rates impacts operating profit by SEK 25.4 million, while EUR/SEK changes affect results by SEK 31.8 million. Management calls it 'natural hedging', investors should recognise it as calculated risk-taking with shareholder money. This unhedged exposure amplifies returns during favourable periods but can dramatically impact results, as witnessed in first-quarter 2025 when currency headwinds reduced operating profit by SEK 20 million.

Geographic revenue distribution shows 40.7% from the United States, creating both opportunity and concentration risk that sophisticated investors cannot ignore. The apparent market penetration success in America, the world's largest radiation therapy market, creates vulnerability to healthcare policy changes, trade tensions, or regulatory shifts that could impact medical device procurement. This geographic mix also explains the currency sensitivity, as dollar and euro revenues convert to Swedish krona, amplifying the currency exposure.

RaySearch added 94 new cancer centers during 2024, expanding to 1,101 clinics globally, impressive growth that nonetheless represents just 13% penetration of the estimated 8,500 radiation therapy centers worldwide. It's quite obvious that substantial expansion runway remains, particularly as existing partnership agreements with equipment manufacturers like Varian provide natural distribution channels. The encouraging aspect: each new customer represents not just immediate license revenue but multi-year support contract potential, compounding the subscription-like economics that make this business model compelling for long-term investors.

To recap,

• Support contracts grew 17.3% and represent 40% of revenue - this recurring stream is the real business, not one-time licenses

• Company capitalises 71% of R&D expenses as "assets" - real cash margins are 5-6%, not the reported 21.9%

• Cash flow (SEK 485M) nearly doubles operating profit because hospitals prepay for multi-year contracts - customers are financing RaySearch's growth

• Completely unhedged on currency with SEK 57M annual profit at risk - each 10% forex move swings results by SEK 25-31M

• Only 13% market penetration despite 1,101 clinics globally - massive runway in an 8,500-center addressable market

People

Walk through RaySearch's Stockholm headquarters and you'll encounter more doctoral degrees per square meter than most university physics departments. These aren't coders debugging features, they're former academics who could be publishing papers on quantum field theory but instead spend their days calculating how to destroy cancer cells without killing patients. 19 percent of the R&D staff hold PhDs, a concentration so extraordinary it explains both the innovation capability and the perpetual tension between academic perfectionism and commercial deadlines.

Johan Löf still controls this empire of algorithms through a mathematical sleight of hand that would make any founder envious. His 52.8% voting control despite owning just 16.1% of shares is a … grief-driven mission protection. The dual-class structure gives his Class A shares ten votes each while ordinary investors get one, ensuring nobody can derail the crusade that began in 1986. When he sold 2 million shares in March 2025, he'd done the math first: converting Class A to Class B shares, selling exactly the right amount to diversify his wealth while retaining 40.5% voting control. Even his personal portfolio management reads like an optimisation algorithm.

The rest of the ownership tells the story of Swedish engineering culture meeting global capital ambitions. Anders Brahme, Löf's former PhD supervisor and co-founder, retains 11.3% voting control, ensuring academic rigour remains embedded in corporate governance. What makes me pause is that foreign ownership has grown to 40.6%, indicating international confidence in a company where three shareholders still control three-quarters of all votes. Cannot say this is democracy; it's basically enlightened despotism with global investors betting their money on Swedish intellectual superiority.

The employees present an even more interesting psychological puzzle. Why do senior technical managers with 20+ years stay for modest equity compensation when Sweden's tech boom offers Silicon Valley-style riches elsewhere? Henrik Friberger and Kjell Eriksson each own only 11,000-12,000 shares despite decades building this company. Deputy CEO Björn Hårdemark literally wrote his thesis here in 2002 and worked his way up through research engineer, physicist, department head, yet holds minimal equity. This isn't the stock-option-heavy compensation typical of technology companies; it's salary-based remuneration suggesting people work here for reasons beyond wealth optimisation.

The answer lies in what happens at 2 AM when a medical physicist reviews treatment plans, knowing patients' lives depend on calculations she cannot independently verify. RaySearch has created something rarer than unicorn valuations: work that matters existentially. When Memorial Sloan Kettering's physics team stays late perfecting dose distributions, it’s not like they are optimising conversion funnels. They're ensuring tomorrow's cancer patients receive precisely targeted radiation that destroys tumours while sparing healthy tissue. That psychological weight creates loyalty no equity package can manufacture.

But the employee satisfaction metrics hint at underlying tensions within this academic-commercial hybrid. The Employee Net Promoter Score of +14 fell short of the >+20 target, while compliance training completion lagged at 87%. These aren't terrible numbers, but they suggest PhD physicists chafing against corporate processes in an organisation that operates more like CERN with a sales team. The cognitive dissonance would break most organisations: brilliant researchers publishing algorithms in peer-reviewed journals while simultaneously optimising for quarterly revenue targets.

Geographic concentration reveals both strategic genius and hidden vulnerability. Seventy-one percent of employees work in Sweden, primarily at Stockholm headquarters where all software development occurs. This hub-and-spoke model attracts global talent, 35 nationalities among 422 employees suggests English has become the working language. But this creates geographic fragility: if Swedish operations were disrupted, core innovation would cease immediately. Every algorithm protecting cancer patients worldwide depends on what happens in one Stockholm office building.

The management dynamics read like an academic department that accidentally became profitable. When analysts challenge Löf about RayCare adoption barriers during earnings calls, his response reveals someone who built something personal being questioned about its quality: "That is simply not true. I don't know where you get that information from." This isn't smooth corporate-speak, is it?

At 86, Hans Wigzell still chairs the board with deep Karolinska connections, the same institution where Löf completed his PhD. Not exactly ceremonial governance theatre; Löf credits Wigzell as crucial during early negotiations when "young PhD students needed a senior voice." The board met fifteen times during 2023, relatively light oversight for a public company generating SEK 1.2 billion annually, suggesting either exceptional management trust or limited formal engagement. Their deviations from Swedish corporate governance codes reflect concentrated ownership making elaborate democratic processes redundant.

The customer relationships extend far beyond typical vendor-client arrangements into genuine research partnerships that create switching barriers beyond technical lock-in. When MD Anderson co-develops liver ablation modules or Memorial Sloan Kettering collaborates on algorithm integration, they're not just buying software, they're investing intellectual capital in shared innovation. These institutions see RaySearch as an extension of their own research capabilities, which explains why 43 of 46 US and Canadian proton therapy centers use RayStation despite premium pricing.

Yet customer decision-making reveals persistent friction between technical excellence and operational reality. Medical physicists appreciate RaySearch's algorithmic superiority, computation times measured in seconds versus minutes with competing systems, but hospital administrators worry about integration complexity and training costs. The result: enthusiastic technical evaluations often yield cautious purchasing decisions as "good enough" bundled solutions from equipment manufacturers win on operational simplicity despite inferior capabilities.

The compensation structure is another chapter of the mission versus money saga. Löf's SEK 13.6 million total compensation ties directly to company performance through a formula paying exactly 2% of profit before tax, no discretionary bonuses, no complex equity schemes. Senior technical staff earn well but modestly relative to their contributions and market alternatives. This is Swedish engineering philosophy applied to human resources.

What we get is a company culture optimised for problems that matter rather than problems that scale. RaySearch employs brilliant people to solve impossible challenges for customers who literally cannot afford mistakes. The human reality drives everything else: concentrated ownership protecting long-term thinking, premium pricing funding continuous innovation, and customer loyalty emerging when software becomes integral to saving lives.

The sustainability question haunts every interaction: Can mission-driven culture survive commercial success? Can academic perfectionism coexist with public company pressures? Can Swedish engineering values scale globally while maintaining their essential character? The answers will determine whether RaySearch's human capital advantage proves durable or becomes its most vulnerable dependency as competitors inevitably target the mathematical moats that currently protect this unique marriage of grief, brilliance, and Swedish methodical precision.

Competition & the Moat(?)

RaySearch sells the one thing their competitors give away for free, and somehow charges premium prices for it. Imagine Tesla competing against car dealers who bundled free Autopilot software with every Honda Accord purchase. That's the upside-down economics Löf's algorithms face daily. Varian, now owned by Siemens Healthineers, dominates through Eclipse software that comes essentially free with their $3 million linear accelerators. Elekta follows identical logic with Monaco systems, treatment planning as digital loss leader to sell hardware. What is not obvious is that this creates the strangest competitive dynamic in enterprise software: David versus Goliath, except Goliath deliberately underprices the very product David sells.

Well, think of Varian as the boxing heavyweight who wins by clinching, not punching. They bundle Eclipse software with radiation machines the way banks once threw in free toasters with checking accounts, except these toasters cost hospitals $200,000 if purchased separately. Being acquired by Siemens brings both deep pockets and divided attention. While Siemens spreads engineering resources across MRI machines, X-ray systems, and robotics, their treatment planning software becomes an afterthought funded by accelerator profits. The consequence is "good enough" algorithms subsidised by hardware margins, creating artificial price competition that pure software companies shouldn't be able to survive.

Elekta presents an interesting strategic puzzle, an independent company squeezed between software specialists and integrated giants. I didn't realise how, let’s say, desperate their position was until I saw their acquisition of Philips' Pinnacle patents. Buying patents rather than developing capabilities signals recognition of software weakness they cannot organically address.

The MIM Software partnership through GE HealthCare is more of the same: they need external expertise to compete with algorithms developed in Stockholm. Given the scale disadvantage against Siemens and technical deficit versus RaySearch, Elekta's survival depends on threading an increasingly narrow needle.

As we mentioned earlier, those 207 PhD-level minds concentrate on radiation therapy optimisation while competitors treat planning software as procurement convenience. This focused approach enables breakthrough capabilities, but creates commercial challenges.

It is tempting to think superior technology automatically wins, until you observe actual procurement dynamics. Let’s take a community hospital in Ohio purchasing its first linear accelerator. The administrator sees Varian's package deal: hmm, a $3 million machine with "free" Eclipse software versus $3.2 million for the same machine plus separate RayStation licensing. The result appears obvious until complexity enters the equation.

In contrast to routine procedures, sophisticated treatments reveal where bundling economics break down. When Memorial Sloan Kettering invests $200 million in proton therapy capability, they won't accept inferior planning algorithms to save $200,000. That suggests market segmentation along sophistication lines: community hospitals choose convenience, academic centres choose capability. The segmentation becomes clearer.

What we are really saying is that RaySearch has achieved something elegant in competitive strategy, they've made it impossible for larger competitors to respond without destroying themselves. Varian and Elekta literally cannot offer truly vendor-neutral software without undermining the hardware lock-in that funds their entire business model. This is how the counter-positioning moat works: competitors face impossible choices between maintaining closed ecosystems that justify premium accelerator pricing or offering open alternatives that commoditise their hardware.

Reading between the lines, Elekta's refusal to provide RayCare interoperability seemed like strategic necessity. Opening their systems would eliminate the switching costs that keep customers buying Elekta machines. Varian only certified RayCare compatibility after customer pressure, then delayed updates to maintain friction. Hmm.

These aren't typical enterprise switching costs measured in training expenses and data migration headaches. When Dr Martinez reviews treatment plans at 11 PM, she's checking calculations, but much more importantly to her, she's protecting her medical licence, institutional accreditation, and patient safety outcomes that extend far beyond software preferences.

Treatment protocols, quality assurance procedures, and regulatory validations all crystallise around specific interfaces that took months to perfect. Most alarming lately, active cancer patients have treatment plans stored in current systems, preventing mid-treatment transitions that could compromise care continuity.

Once medical physicists spend six months learning RayStation workflows, they become intellectually invested in approaches that cannot transfer to competing platforms.

As noted earlier, 49% of RaySearch employees work in R&D, creating innovation density that diversified competitors cannot match through divided attention. This is focused capability versus scattered resources: while Siemens Healthineers spreads engineering across imaging, robotics, and diagnostics, RaySearch concentrates doctoral-level minds on radiation therapy optimisation. Each innovation cycle widens the gap between specialist algorithms and bundled alternatives.

Probably the most revealing competitive insight comes from geographic variation. RaySearch achieves 50% market share in UK and Benelux markets where open procurement processes favour technical evaluation over bundling convenience. European healthcare systems, with their emphasis on long-term operational efficiency, naturally select superior algorithms despite higher initial costs. In contrast to European markets, US dynamics favour integrated vendors through private ownership models that emphasise initial capital efficiency over operational optimisation.

As we mentioned earlier, RaySearch commands 85% market share in proton therapy, not just dominance but mathematical fortress-building. Treatment complexity that exceeds "good enough" algorithms destroys bundling advantages completely. When centres invest $200 million in proton capability, they cannot accept planning software limitations that compromise precision worth tens of millions in infrastructure investment.

I was thinking about why vendor neutrality might actually be a strength rather than weakness, after all, RaySearch lacks the hardware relationships that drive competitor revenue. When looking closer though, you see that working with equipment from IBA, Accuray, Hitachi, Varian, and Elekta creates compatibility advantages that competing software cannot match without undermining parent company strategies. When cancer centres operate mixed equipment fleets, vendor-neutral planning becomes essential rather than convenient.

The company hasn't lost significant moats over time, if anything, competitive advantages have strengthened as treatment complexity increases.

Mr. Market

Mr. Market recognised a fundamental business model transformation: RaySearch evolved from a marginally profitable software vendor into a high-margin, recurring revenue machine with defensible moats.

The result was a 350% appreciation from SEK 68 (December 2022) to SEK 306 (May 2025) as investors repriced the company from cyclical healthcare supplier to scalable software business with subscription-like economics.

The inflection point arrived in Q3 2023 when operating margins hit 11% on SEK 253 million sales, proving the operational leverage we described earlier had become reality, not aspiration. Mr. Market responded by driving the stock from SEK 72 to SEK 90 through year-end, a seemingly modest 25% gain that marked the beginning of systematic re-rating. Each subsequent quarter validated the thesis: Q1 2024's 17.8% margins pushed shares to SEK 127, Q2's 25% margins drove prices to SEK 157, and Q3's sustained 21% margins reached SEK 195 before peaking at SEK 216 in December.

The March 2025 founder sale created the ultimate market confidence test. When Johan Löf dumped 2 million shares at SEK 250, liquidating 5.8% of the company, sophisticated investors treated it as diversification, not distress. The stock's recovery to SEK 306 by May following Q1 2025's 29% revenue growth suggests Mr. Market views the competitive moats and switching costs we examined as genuinely structural, not dependent on founder presence.

Current pricing embeds Silicon Valley-level growth expectations for a Swedish niche software company. At 44x earnings and 7.3x sales, RaySearch trades at premiums typically reserved for hypergrowth SaaS companies, despite serving just 13% of global radiation therapy centers. The SEK 10.4 billion market capitalisation assumes sustained 25%+ revenue growth and expanding margins, mathematical perfection that leaves little room for execution disappointments.

The embedded expectations reveal Mr. Market's dual conviction and blind spots. Currency exposure that management refuses to hedge creates 20% annual profit volatility, yet investors seemingly ignore this structural risk. Geographic concentration with 40.7% of revenue from the United States, the market we noted earlier, adds another layer of dependency that current valuations treat as irrelevant.

The mathematics are clear: Mr. Market believes the algorithmic moats and customer dependency justify premium pricing, but has priced those advantages to perfection with minimal margin for error.

To recap,

• At 44x earnings and 7.3x sales, the market has priced RaySearch as a high-growth SaaS company despite operating in a specialised hardware-adjacent niche with just 1,100 customers globally.

• The 350% price appreciation from SEK 68 to SEK 306 represents a complete business model re-rating, from cyclical software vendor (15-20x P/E) to recurring revenue subscription business (40-50x P/E).

• Current multiples embed expectations of 25%+ revenue growth and 25%+ operating margins for a company that has achieved these metrics for exactly four consecutive quarters.

• The SEK 10.4 billion market cap implies the market believes 13% penetration of global radiation therapy centers justifies premium software company valuations typically reserved for companies with vastly larger addressable markets.

• Price resilience during the founder's SEK 500 million share sale indicates the market views competitive advantages as structural rather than management-dependent, algorithms and switching costs matter more than founder presence.

Bear Thesis

RaySearch's mathematical precision has created an ironic vulnerability: the very systematic approach that optimises radiation delivery makes the company predictably exposed to systematic disruption. As we noted earlier, their premium positioning depends on serving elite institutions, but that prestigious customer base now faces existential crisis. The NIH's $1.81 billion funding termination across 694 grants in early 2025 is no regular budget tightening; it's the systematic dismantling of the academic medical center model that justifies premium treatment planning software. When Memorial Sloan Kettering loses federal grants, they don't upgrade algorithms, they ration chemotherapy.

The currency exposure we examined becomes particularly dangerous now. That SEK 57 million annual profit at risk from unhedged foreign exchange is akin to a financial roulette during trade war escalation. Management's admission that Q1 2025's SEK 20 million currency loss stemmed from "political turmoil in the US" reveals how Swedish engineering pragmatism meets American geopolitical chaos. With comprehensive tariffs targeting healthcare equipment specifically, each quarterly report becomes a forex lottery ticket where mathematical brilliance can't hedge away political instability.

The working capital dynamics we explored create a particularly cruel trap. Those 142 days of revenue tied up in customer promises become dangerous when economic conditions deteriorate and academic centers delay payments due to funding uncertainty. Growth becomes self-destructive: each new installation requires more working capital investment while existing customers extend payment cycles. The faster RaySearch expands globally, the more cash gets trapped in institutional bureaucracies facing budget crises.

Even more concerning is the SEK 530 million capitalised development asset, 27% of the balance sheet representing years of algorithmic innovation. But what happens when AI democratises sophisticated treatment planning? The same technological advancement RaySearch pursues could make their premium capabilities available through cloud-based alternatives at commodity prices. We mentioned earlier how this creates switching cost advantages, but it also creates obsolescence risk. That half-billion-krona asset could require massive impairments if planning software becomes good enough rather than mathematically perfect.

You would think prestigious customer relationships provide stability... except the academic medical center model faces systematic collapse. When Columbia University loses 157 NIH grants simultaneously, when chemotherapy shortages force treatment rationing, the entire prestige medicine infrastructure starts crumbling. RaySearch built their moats around institutions that justified premium pricing through research missions and clinical excellence. Remove the research funding, add drug shortages, and suddenly "world-class treatment planning" becomes an unaffordable luxury during healthcare austerity.

The 350% stock appreciation we traced from Mr. Market's re-rating assumes these converging headwinds somehow miss their target. At 44x earnings, current pricing embeds mathematical perfection: sustained 25% growth, expanding margins, flawless execution during currency volatility, competitive moats holding against vastly better-resourced opponents, and academic customers maintaining premium purchases despite funding crises. That's a lot of variables requiring simultaneous optimisation, the kind of complex system that brilliant algorithms usually warn us to avoid.

The reckoning approaches through cascading second-order effects: funding cuts trigger customer budget pressure, trade tensions amplify currency volatility, AI advancement commoditises premium features, and institutional healthcare shifts toward "adequate" rather than "excellent."

Johan Löf's grief-driven mission to solve cancer's precision challenge may have created something mathematically beautiful, and financially vulnerable to forces no algorithm can optimise away.

Bull Thesis

The Oracle of oncology is hiding in plain sight.

While investors obsess over ChatGPT and autonomous vehicles, RaySearch has quietly become invisible infrastructure for cancer treatment, the kind of mission-critical software that enterprises pay premium prices to forget exists.

We explored earlier how customers desperately hope this software becomes invisible, but that's precisely what creates platform economics. When 1,100 cancer centers depend on algorithms they cannot independently verify, you're not selling software... you're selling mathematical assurance. The recurring support revenue we examined, 40% of total sales growing 17% annually, represents subscriptions to staying current with life-saving innovations. This is medical advancement insurance.

Complexity becomes competitive advantage in ways that bundling strategies cannot match. As cancer treatment evolves toward precision medicine, genetic profiling, and AI-driven adaptive therapy, the gap between RaySearch's specialised algorithms and competitors' "good enough" bundled solutions becomes unbridgeable.

Those 201 AI segmentation models and seconds-versus-minutes computation speeds we discussed, much more than simply technical improvements, are mathematical moats that widen with every breakthrough. Varian and Elekta face an impossible choice: invest billions developing competitive algorithms or maintain the hardware subsidies that justify current pricing. Meanwhile, RaySearch's 49% R&D workforce concentration creates innovation density that diversified competitors simply cannot replicate without destroying their business models.

The platform transformation is already happening, just unrecognised by most investors. RayCare's certification with Varian TrueBeam systems unlocks access to 7,000 installed machines globally, but the real opportunity extends far beyond interoperability.

When hospitals integrate treatment planning, patient management, and clinical workflows into a single platform, they're buying software modules, sure, but most importantly they're subscribing to the operating system for oncology. The Heyou Hospital's SEK 77 million commitment represents more than China expansion; it validates comprehensive workflow integration worth multiples of standalone planning software. You would think this platform vision would command attention, but investors remain fixated on quarterly license fluctuations while missing the subscription transformation occurring beneath the surface.

The geographic mathematics are staggering once you recognize we're witnessing market creation, not market share battles. That 13% penetration across 8,500 global radiation therapy centers represents a 25-year runway in established markets, but demographic trends suggest something far more compelling. With global cancer cases expected to reach 30 million by 2040, a 55% increase, we're not just talking about existing centers upgrading software. We're talking about thousands of new treatment facilities requiring sophisticated planning capabilities from day one.

The developing world's healthcare infrastructure build-out creates greenfield opportunities where technical superiority matters more than incumbent relationships. When cancer centers invest $50-200 million in equipment, they won't accept inferior planning algorithms to save $200,000 on software.

In an AI-disrupted world, regulatory barriers become competitive advantages rather than bureaucratic obstacles. Those 18-month FDA approval cycles and country-specific clearances we mentioned create time machines that newer competitors simply cannot access. While Silicon Valley engineers debate whether AI will replace radiologists, RaySearch's 25-year regulatory track record provides 2-3 year head starts on any algorithmic breakthrough. The PhD concentration is intellectual firepower, but even more than that, it's institutional knowledge about navigating medical device regulations that takes decades to accumulate. When breakthrough innovations require both algorithmic sophistication and regulatory expertise, the combination becomes nearly impossible for new entrants to replicate quickly.

Mission-critical applications eliminate price sensitivity in ways that conventional enterprise software cannot achieve. When medical physicists review treatment plans at 2 AM, they're not optimising software budgets, they're protecting medical licenses, institutional accreditation, and patient safety outcomes that extend infinitely beyond procurement considerations.

The switching costs we explored transcend training expenses to encompass professional liability, workflow disruption, and treatment continuity for active cancer patients. Once algorithms become integral to saving lives, customers will pay almost any reasonable price to avoid disruption. This creates pricing power that software companies serving less critical functions can only dream about achieving.

The asymmetric optionality becomes clear when you map multiple expansion vectors against downside protection. Geographic expansion, platform integration, AI capabilities, and strategic acquisitions each represent independent paths to value creation, while the recurring revenue base and debt-free balance sheet provide substantial downside cushion.

With SEK 503 million in cash and minimal customer churn, RaySearch could survive extended market disruption while competitors burn through financing. But here's the contrarian insight: this isn't a defensive position masquerading as growth opportunity.

The convergence of demographic trends, treatment complexity, and platform economics suggests multiple scenarios where this becomes a significantly larger business, with relatively few scenarios involving permanent capital impairment. When mission meets mathematics in a growing market with structural switching costs, the asymmetric risk-reward profile becomes compelling for patient capital.

So what do we make of all this?

Here's what strikes me about the radiation therapy business: it's infrastructure you pray never to need, operated by people you'll never meet, funded by institutions increasingly squeezed for cash. Cancer rates climb relentlessly while healthcare budgets contract. The mathematics are brutal.

More patients, same resources, higher precision requirements. Someone has to solve this equation, and it won't be through good intentions alone. It requires exactly the kind of algorithmic precision RaySearch has spent 25 years perfecting. The macro setup creates demand for mathematical solutions while simultaneously pressuring the customers who buy them. Classic healthcare contradiction.

What is RaySearch, really?

It's a grief-driven algorithm that became a business model. Johan Löf channeled personal loss into mathematical precision, then discovered that life-and-death software creates the stickiest customer relationships imaginable. The company's strength is obvious: once you're trusted with radiation dose calculations, switching becomes professionally suicidal.

Medical physicists stake their licenses on these algorithms. But here's the vulnerability everyone misses: RaySearch built their entire castle on the highest ground, elite academic medical centers that justify premium pricing through research missions. Remove NIH funding, add drug shortages, and suddenly "world-class" becomes an unaffordable luxury. They've optimized for customers facing systematic budget pressure.

The investor psychology split is fascinating.

Growth-oriented minds see demographic tailwinds, platform potential, and a 25-year runway across 8,500 global treatment centers. They're buying the Salesforce-for-oncology story.

Value-focused investors see a 44x earnings multiple during systematic headwinds, currency exposure management refuses to hedge, and geographic concentration in exactly the markets facing political chaos.

Same company, different movies. The growth crowd sees switching costs and platform economics. The value crowd sees valuation risk and macro headwinds. Both are right. Both are incomplete. This is simultaneously a defensive moat business and a high-growth software story, which explains why it trades like neither.

Let me think through three plausible scenarios for this business.

First: the platform transition works, RayCare becomes the operating system for oncology, and RaySearch expands from 1,100 customers to 5,000+ as cancer treatment standardises globally.

Second: AI democratises sophisticated planning algorithms, cloud-based alternatives emerge at commodity pricing, and premium positioning erodes gradually then suddenly.

Third: healthcare consolidation accelerates, Siemens and Philips acquire regional competitors, bundled vendors squeeze out specialised players through sheer procurement leverage.

The wild card? Geopolitical fragmentation where Chinese hospitals can't buy Swedish software, American centers can't integrate European systems. Technology nationalism meets cancer treatment. None of these scenarios feel inevitable, but all feel possible.

This company represents something larger than radiation therapy software. It's specialized expertise meeting regulated industries where technical excellence creates sustainable moats. The pattern repeats across healthcare, financial services, aerospace: concentrated PhD talent, regulatory barriers, customer lock-in through professional liability.

These businesses share common DNA, brilliant people solving impossible problems for customers who literally cannot afford mistakes. They trade at premiums or discounts depending on whether markets recognize the switching cost dynamics. Most investors never encounter them because they're too specialized, too niche, too far from Silicon Valley's venture narrative. But they often represent the actual infrastructure making modern life possible.

What happens when intensely personal missions encounter quarterly earnings pressure? Johan Löf's story isn't just about algorithms and cancer treatment. It's about whether grief-driven excellence can survive public market scrutiny, whether Swedish engineering values scale globally, whether mathematical precision translates to financial durability.

The deeper question haunts every mission-driven company: can you serve two masters simultaneously, the problem you're solving and the shareholders funding the solution? RaySearch suggests the answer might be yes, but only if the mathematics work perfectly. And mathematics, unlike markets, rarely account for the human variables that ultimately determine whether brilliant algorithms become enduring businesses or cautionary tales about optimising for the wrong variables.

If I had to recap in a few sentences,

• The most important infrastructure is the kind you pray never to use personally

• Mission-driven businesses face an impossible equation: serve the problem or serve the shareholders, rarely both perfectly

• When your customers literally cannot afford mistakes, switching costs transcend economics and become existential

• The same company can simultaneously be a defensive moat story and a high-growth platform play, which is why markets often misprice specialised B2B software

• Grief-driven algorithms that save lives create stickier customer relationships than any SaaS retention metric could measure

---

LEGAL DISCLAIMER AND DISCLOSURE

© 2025 Silba. All rights reserved.

This analysis is for informational purposes only. The content represents the views and opinions of the author, based on publicly available information. While we strive for accuracy, we make no representations or warranties regarding the completeness or accuracy of the information presented. Errors will occur.

This analysis is not investment advice. It is not a recommendation to buy, sell, or hold any security. The author may hold positions in securities mentioned. Future performance of any investment discussed cannot be guaranteed.

All financial figures, unless otherwise noted, are approximations based on publicly available information. Readers should verify all claims, statistics, and analytical processes independently. The information contained herein is not intended to be a complete analysis of any security, company, or industry mentioned.

Market conditions change. Past performance is not indicative of future results. Investments mentioned may not be suitable for all investors. Readers should conduct their own due diligence and consult with professional financial, legal, and tax advisors before making any investment decisions.

By accessing this analysis, you agree that neither Silba nor its affiliates bear any responsibility for investment decisions you make based on this information.