This is investment research, not personal financial advice.

The rally and the removal

Pro Medicus (ASX: PME) closed at A$177.20 on 24 June 2026 (ASX 2026), more than 60% above its 14 May low near A$107.75 and about 36% higher over the past month (Stockanalysis 2026) — a violent re-rating powered by a run of US hospital contract wins for its Visage medical-imaging software. After touching A$178.99 the prior session, the stock eased into the close on a day dominated not by a fresh deal but by index mechanics: the company is being removed from the S&P/ASX 50, and passive funds were rebalancing out of it (Simply Wall St 2026). The same session, a widely-read valuation note argued PME was still 11.3% undervalued — at 77 times earnings.

That juxtaposition is the story. In six weeks the market went from treating Pro Medicus as an AI-disruption casualty — the stock was down roughly 40% for 2026 by mid-May — to bidding it back up on signed, multi-year contracts, even as the company drops out of the index that helped anchor its institutional ownership. At A$177.20 the market capitalisation is about A$18,512 million — roughly A$18.5 billion (ASX 2026) — against FY2025 net profit of A$115.2 million. The question the move forces is the one PME has provoked for years, only sharper after a 60% bounce: does the price reflect newly contracted growth, or a valuation that prices a decade of perfection?

The honest answer is both, and the distinction matters. The contracts are real and the growth is contracted. The level is not a function of this quarter's wins; it is a function of how many more years of 20%-plus compounding the buyer is underwriting. This article builds the value from the business drivers and then places the post-rally price against it.

The proximate catalysts are concrete. Since mid-May, Visage Imaging — Pro Medicus's wholly-owned US subsidiary — has signed a seven-year, A$90 million agreement with Boston's Beth Israel Lahey Health, a five-year, A$28 million renewal-and-upgrade with Pittsburgh's Allegheny Health Network, and a five-year, A$16 million renewal with Ohio State's Wexner Medical Center that added Visage 7 Workflow and Cardiology (Visage Imaging 2026; Capital Brief 2026). On the Allegheny renewal, chief executive Sam Hupert made the bull case explicit, arguing the wins "tend to disprove the theory that all software companies will be negatively disrupted by AI" (Capital Brief 2026).

What Visage actually sells

Pro Medicus licenses Visage 7, an enterprise imaging platform that radiologists use to view, report on and archive diagnostic images — the software layer between a hospital's scanners and the doctor reading the study. Its technical claim is architectural: a single server-side rendering code base streams full diagnostic studies in seconds over standard networks, rather than pushing entire image sets to a local workstation. The same code base now carries cardiology and digital pathology, which is how the company sells "full stack plus one" deals into accounts it already owns.

The economics flow from three choices. First, the product is sold mostly to large US academic medical centres and integrated delivery networks (IDNs) — high-prestige, high-volume reference accounts whose adoption pulls the next buyer along. Second, contracts are long (five to ten years) and increasingly priced on transaction volumes, so revenue grows with imaging activity even between new wins. Third, almost everything is delivered as software: Pro Medicus writes the code, the customer runs the exams, and the marginal cost of another study is close to nothing. That is why gross margin sits at 99.9% (Pro Medicus 2025) — a figure that looks like a typo until you realise the company books almost no cost of goods because there are almost none.

The compounding engine, then, is not a single big sale; it is a reinforcing loop. A marquee win (Beth Israel Lahey, Allegheny, Ohio State, and last year's University of Colorado deal) validates the platform, which lowers the sales friction for the next academic centre, which adds transaction revenue that funds R&D into new modules (cardiology, pathology, AI orchestration), which deepens switching costs at the installed base. The loop has run for a decade and shows up cleanly in the numbers.

A near-100% gross margin, and what it hides

FY (to 30 June) Revenue A$m NPAT A$m EPS A$ Gross margin FCF A$m Net cash A$m ROIC*
2021 68.1 30.9 0.29 99.3% 39 62 37%
2022 93.5 44.4 0.42 99.5% 61 88 41%
2023 124.9 60.7 0.58 99.6% 62 120 42%
2024 161.5 82.8 0.79 99.8% 82 153 42%
2025 213.0 115.2 1.10 99.9% 111 208 43%

*Author-computed ROIC; see the bridge below. Source-reported figures: Pro Medicus 2025, Pro Medicus 2024, Pro Medicus 2023, Pro Medicus 2022; multi-year series cross-checked against Stockanalysis 2026. "Net cash" is the negative of net debt (the company is debt-free).

Revenue compounded at roughly 33% a year over the five years, net profit at about 39%, and earnings per share quadrupled from A$0.29 to A$1.10. The mix improved as it scaled: underlying EBIT margin reached about 75% in FY2025, and free cash flow of A$111 million was 96% of net profit — accrual earnings that turn almost entirely into cash.

Returns on capital are the part that needs care. Measured as author-computed ROIC — net operating profit after tax over total capital employed (equity plus the trivial debt) — Pro Medicus earns about 43% (FY2025), and the series has sat in a stable high-30s-to-low-40s band for years. But that understates the true return on the operating assets, because more than A$200 million of the capital base is idle cash. Strip the surplus cash out and the return on the capital actually used to run the business exceeds 200%. The economically meaningful read is that growth needs almost no incremental capital: on an operating basis, author-computed incremental ROIC has run above 150%, because each new dollar of revenue arrives at near-incremental gross margin and the company funds its R&D and sales expansion from operating cash flow. The flip side — and it matters for capital allocation — is that the business cannot reinvest most of what it earns, so cash piles up.

The cash machine and the one-off that flattered the multiple

The balance sheet is the opposite of a risk. Pro Medicus held A$221.8 million of cash and investments and no debt at 31 December 2025 (Pro Medicus 2026); full-year FY2025 closed with about A$208 million of net cash (Pro Medicus 2025). Owner earnings track free cash flow closely: FY2025 operating cash flow of roughly A$120 million, minimal maintenance capex, negligible stock-based compensation and little capitalised development leave owner earnings near the A$111 million of reported free cash flow. There is no working-capital drain (customers pay; there is no inventory), no acquisition goodwill of consequence, and no refinancing wall. On survivability, this is as close to bulletproof as an ASX growth name gets.

That cash quality also exposes a trap in the headline multiple. Several data providers quote a trailing price/earnings ratio near 77-84x (Simply Wall St 2026; Stockanalysis 2026), which sounds extreme but not absurd for a 30%-plus grower. The catch is that the trailing twelve months include a large non-operating gain. Pro Medicus reported statutory first-half FY2026 net profit of A$171.2 million, up 230.9% — but underlying EBIT rose 29.7% to A$90.7 million at a 72.6% margin (Pro Medicus 2026). The gap is a one-off, chiefly investment and revaluation gains rather than software profit; the underlying half grew about 30%, in line with the operating line. So the "77x trailing" multiple is calculated on earnings inflated by a one-off. On the audited FY2025 result (EPS A$1.10), the 24 June price is about 160 times earnings; on a consensus FY2026 underlying figure near A$1.55-1.60 a share (author estimate), it is roughly 110-115 times. The cheaper-looking number is the one to trust least.

The moat: switching costs, speed, and the AI question

The moat is real and shows up in a number bears rarely engage with: a 100% contract renewal rate, with every renewal struck at a higher price point than the original (Pro Medicus 2025). Once Visage 7 is the reporting environment a radiology department lives in, ripping it out means retraining clinicians, re-validating workflows and risking diagnostic throughput — switching costs that are operational, not just contractual. The Ohio State and Allegheny renewals, both taken at higher per-transaction economics with added modules, are the moat working in real time (Visage Imaging 2026; Capital Brief 2026). The architecture advantage compounds it: because cardiology and pathology run on the same code base, Pro Medicus can cross-sell new clinical domains into existing accounts at low incremental cost, widening the moat rather than merely defending it.

The counter-evidence deserves equal weight, because it is what drove the 40% drawdown earlier in 2026. The bear thesis is that generative AI and foundation-model viewers could commoditise the diagnostic-viewing layer; that hyperscalers (AWS HealthImaging, Google) and the modality incumbents (GE, Philips) could bundle cloud imaging and undercut a A$170 million seven-year deal; that customer concentration in elite US centres makes the order book lumpy; and that ever-larger contracts will eventually meet procurement resistance. Founders Sam Hupert and Anthony Hall still control roughly half the register, and periodic founder share sales feed the "insiders are taking money off the table" narrative.

The evidence cuts mostly the other way, for now. Radiology is the overwhelmingly dominant field for medical AI — about 74% of more than 1,350 AI/ML-enabled devices authorised by the US FDA (FDA 2026) — but those algorithms still need a workflow and viewer to be deployed at the point of care, which is precisely the orchestration layer Visage occupies. In that framing AI is a tailwind PME hosts, not a wave that drowns it. The structural risk is genuine but unproven: a foundation-model viewer that is "good enough" and radically cheaper would erode the speed advantage that justifies the price. That is the bear case to take seriously, and it is unresolved.

What A$18.5 billion is pricing

A capital-light, net-cash software compounder is best valued on owner earnings and a multi-stage discounted cash flow, cross-checked with a reverse DCF — not on book value or a peer P/E, because there is no comparable asset intensity and the peer set is far slower-growing. Veeva Systems, the closest listed analogue (a vertical healthcare-software franchise with high margins), trades around 27 times trailing and 31 times forward earnings (Veeva 2026); the global healthcare-services average is about 26 times (Simply Wall St 2026). Even the most expensive ASX software names sit far below PME: WiseTech around 53x trailing and TechnologyOne around 62x. Pro Medicus at ~110-160x is in a category of one.

Build the value from drivers. Start from FY2025 owner earnings of about A$111 million, net cash of A$208 million, and a discount rate of 9.5% — deliberately modest, because a long-duration growth stock's value is acutely sensitive to the rate, and a higher cost of capital would cut these figures materially. A base case of ~20% revenue growth for seven to eight years, fading to ~8% terminal, holding the ~70% EBIT margin and exiting at ~30x owner earnings, supports roughly A$150-195 a share. The current A$177.20 sits at the upper end of that band — meaning the market is already underwriting most of a base case that itself assumes near-flawless execution for the better part of a decade.

The reverse DCF makes the wager explicit. At A$177.20 the enterprise value (market cap less net cash) is about A$18.3 billion, or roughly 165 times FY2025 free cash flow. To justify that on a 9.5% discount and a 25-30x exit, owner earnings must compound at roughly 22-25% a year for a decade and the premium multiple must survive. That is the literal meaning of "pricing years of perfection": not that the business is bad — it is exceptional — but that the price has already spent most of the exceptionalism.

Scenario Key driver Value/share
Severe downside AI/cloud viewers erode the platform; growth fades to ~10%; multiple resets to ~30-40x A$55-85
Bear growth slows to mid-teens; margins steady; multiple compresses to ~45-55x A$95-130
Base ~20% CAGR 7-8 yrs fading to ~8%; ~70% EBIT margin; 9.5% discount; ~30x exit A$150-195
Bull ~25-30% CAGR for a decade on cardiology/pathology/AI optionality; premium sustained A$230-320

A sensitivity check shows the two variables that move everything: the duration of 20%-plus growth and the exit multiple (a proxy for the discount rate and terminal economics). Keep the base-case growth path but compress the exit multiple from 30x to 20x, and value falls toward the bear range; extend 20%-plus growth from eight years to twelve at a sustained multiple, and value pushes into the bull range. Revenue growth and the terminal multiple, not this year's margin, are the swing factors.

Bull, bear and the crux

The reaction verdict: the rally is roughly proportionate to the information. A string of signed, multi-year, higher-priced contracts is genuine new evidence, and the earlier 40% drawdown had priced an AI-disruption scenario the contract wins directly contradict — so a sharp re-rating off an oversold base is defensible, and the S&P/ASX 50 removal is a passive-flow event, not a fundamental one. What the bounce does not do is make A$18.5 billion cheap. The price sits at the top of a base case that assumes a decade of ~20% compounding; the asymmetry from here is multiple compression on any growth or margin wobble, not business collapse. Pro Medicus can keep executing superbly and still disappoint a A$177 share price if growth normalises a few years early.

The bull case requires the compounding engine to keep widening: cardiology and pathology scaling into the installed base, transaction revenue lifting as imaging volumes grow, US enterprise penetration accelerating, and Visage positioned as the platform that orchestrates FDA-cleared AI rather than the product it replaces. The bear case requires only that growth duration disappoints — that the pool of elite US accounts is finite, deals get lumpier, and the multiple normalises toward even a generous 50x.

The crux is three questions, each with a clock. Does ~20%-plus revenue growth persist as the largest accounts are penetrated and the model leans on transaction volumes and new modules — answered progressively at the FY2026 result on 14 August 2026 and across the FY2027-FY2028 contract cadence? Is radiology AI a tailwind Visage hosts or a foundation-model wave that commoditises the viewer — a two-to-four-year question as authorised algorithms move from clearance to deployment (FDA 2026)? And is the index removal a temporary overhang or a marker that the re-rating has outrun the free float — resolved at the June 2026 rebalance and subsequent reviews?

What the next results must show

These are observations about what to watch, not instructions. The first read is the cadence and size of new full-stack wins: the base case needs roughly five or more a year at stable-to-rising contract values; a slowdown there erodes the growth-duration assumption the price depends on. The second is the underlying EBIT margin — sustained below ~68% would test the operating-leverage story. The third is transaction-linked revenue: if volume-based revenue decelerates while licence revenue flattens, the cloud compounding engine is maturing earlier than the multiple assumes. The fourth is renewal quality: the 100% renewal rate at higher prices is the cleanest moat signal there is, so any flat-priced renewal or non-renewal is material. The last is earnings quality and capital allocation — whether intangibles start rising faster than revenue (capitalised development flattering profit) and whether founder selling continues, given Hupert and Hall still own about half the company and the cash pile keeps building with no obvious high-return home.

Source notes and confidence

Confidence is high on the multi-year financial history, which comes from Pro Medicus's FY2025 financial report and prior annual reports (Pro Medicus 2025; Pro Medicus 2024; Pro Medicus 2023; Pro Medicus 2022), with the series cross-checked against an aggregator (Stockanalysis 2026); verification is set to partial because figures were read from results releases and a compiled history rather than each PDF being re-fetched line by line. ROIC, incremental ROIC and owner earnings are author-computed from those inputs and labelled as such; the ROIC band is presented on a total-capital-employed basis, and the >200% operating-capital and >150% incremental figures are flagged as distorted upward by the company's net-cash, asset-light structure. The market snapshot — A$177.20 and an A$18,512 million market capitalisation — is the 24 June 2026 ASX close (ASX 2026). The FY2026 forward EPS used in the multiple is an author estimate. The first-half FY2026 statutory profit of A$171.2 million includes a large one-off gain whose exact composition is not fully disaggregated here; the underlying ~30% growth is the operating signal (Pro Medicus 2026). Contract values are minimum-commitment figures from company releases (Visage Imaging 2026; Capital Brief 2026); actual revenue can run higher under the transaction model. Customer-level concentration and segment economics are only partly disclosed. Market-size and AI-adoption context is third-party estimate (Mordor Intelligence 2026; FDA 2026), and peer multiples are point-in-time (Veeva 2026). The 11.3%-undervalued claim cited in the lede is one external model (Simply Wall St 2026), included as market context, not as this article's conclusion.

References