INFINITT Healthcare(071200) _ A Hospital Software Company Whose Balance Sheet Looks Like a Fund
A Hospital Software Company Whose Balance Sheet Looks Like a Fund
INFINITT Healthcare Co., Ltd. · KOSDAQ1 071200 — Operating profit from the core business grew 3.6-fold over three years, and about two-thirds of what the company earned over the same period sits in overseas exchange-traded funds (ETFs). This report reads the operating business and the capital allocation as two separate things.
Disclaimer · Please read this first
This report is not investment advice and is not a buy or sell opinion on any security. Responsibility for any investment decision and its outcome rests with the reader, and loss of principal is possible.
The author is not a financial institution, a registered investment adviser, or an unregistered advisory service. This is an analysis published for the author's own study and record-keeping.
The reference date is 1 September 2026. Disclosures or facts that changed after that date are not reflected.
All figures are taken from original regulatory filings. Estimates contain the author's own assumptions, and their accuracy or completeness cannot be guaranteed. Where this report differs from the original filing, the filing is correct.
The author did not look at the market's short-term valuation first. The order of work was to check the competitiveness of the core business and the direction of earnings first.
Holdings: as of the reference date the author does not hold this security. No sponsorship, writing fee, or IR-agency payment was received. That said, the author may buy or sell this security after publication without prior notice.
Five key points
The owner of net income is not the operating business but an overseas index.
Of H1 2026 net income of KRW 38,740 mn (USD 27.7 mn), financial income was KRW 30,060 mn (USD 21.5 mn), of which ETF valuation gains were KRW 28,570 mn (USD 20.4 mn). ETF disposals over the same period were KRW 110 mn (USD 0.08 mn) — this gain has never become cash. The company itself disclosed that a 5% move in fair value swings net income by ±KRW 9,890 mn (USD 7.1 mn), and a 10% move in exchange rates by ±KRW 21,000 mn (USD 15.0 mn).
Fact · Semi-annual report notes 11 and 33, II-5
The improvement in the core business is real, but about half of it came from costs.
First-half gross profit rose from KRW 31,430 mn to KRW 38,960 mn, up KRW 7,530 mn (USD 5.4 mn). At the same time selling, general and administrative expenses (SG&A) fell from KRW 26,530 mn to KRW 24,570 mn. Most of the decrease sits in two lines — bad debt expense (KRW 1,820 mn → 200 mn) and fees paid (KRW 4,370 mn → 2,730 mn) — and both returned to their 2024 levels. Rather than a one-off item disappearing, this reads as costs returning to their earlier level, but that is an [Interpretation].
Fact · note 28
What keeps competitors out is not technology but the cost of switching.
This software runs 24 hours a day, 365 days a year, and if it stops, image reading stops. It is tied by interfaces to the electronic medical record (EMR) and hospital information system (HIS), and replacing it means migrating petabytes of data all over again. The result shows up in revenue — H1 2026 revenue "transferred over time" was KRW 37,320 mn (USD 26.7 mn), or 64.8% of sales, up 23.8% year on year. The technical standard (DICOM) is open, so it does not block entry by itself; pulling out a system that is already installed is a separate matter.
Fact · notes 20 and 38 · company description in filings
For three years, growth has come only from outside Korea.
From FY2023 to FY2025, revenue by region changed as follows: Korea −0.7%, Americas +43.4%, Europe +35.2%, Asia +25.7%. In H1 2026, Europe led with +47.4%. However, the Americas — the largest region by revenue — grew only 6.2%, roughly at market speed, and among the overseas entities where growth occurred, North America (equity −KRW 2,400 mn) and the Middle East & Africa (−KRW 5,290 mn) are in a capital deficit. Where the growth is and where the capital is have come apart.
Fact · notes 38, 11-12
The domestic market share figure is a 2019 number.
The 75% share of tertiary general hospitals2 and 70% of general hospitals that the company states in its periodic reports are based on hospitals registered with HIRA3 as of 2019. Since then, press reports confirm that Taeyoung Soft has won new or replacement projects at 7 of the 47 tertiary general hospitals in the 5th designation cycle. No public source exists to confirm the current figure.
Fact (2019 basis) + press
Sources · Consolidated financial statements (annual report filed 20 March 2026 [amended] · semi-annual report filed 14 August 2026) · FY2021–22 from the original pre-restatement filing (FY2022 annual report)
The company and its products
What it makes and who it sells to
This company makes and sells software that stores and transmits the medical images hospitals capture.
It was founded on 30 December 2002 and listed on KOSDAQ on 26 May 2010. Its head office is in Guro-gu, Seoul, and its fiscal year ends in December. It has no factory — the company states in its periodic report that it "has no plant-level production facilities, production output, or utilisation rate," and its tangible assets are office space and fixtures attached to its R&D centre. Raw materials (finished goods such as servers and PCs) account for 5.8% of cost of sales. Headcount was 242 as of H1 2026, of whom 119 are in R&D (122 in 2023).
There are 11 consolidated subsidiaries, all unlisted — eight overseas (North America, Taiwan, Japan, Shanghai, Europe, Southeast Asia, Middle East & Africa, Brazil) and three domestic (INFINITT Ventures, Zeron Healthcare, Techheim). The North American entity, the largest overseas, began in 1999 as a sales distributor for the US market and became a subsidiary in 2009 when the company acquired a 98% stake.
The corporate credit rating from 한국평가데이터, a Korean commercial credit information agency, rose two years running: A (2024) → A+ (2025) → AA− (23 April 2026). That rating measures commercial creditworthiness, not business competitiveness or the share that reaches shareholders. [Fact · semi-annual report I-1]
| Ownership | Value |
|---|---|
| Largest shareholder | Solborn Inc. 48.76% · 51.17% including related parties (30 June 2026) |
| Subsequent change | Largest-shareholder side 51.80% (10 July 2026 large-holding report) |
| One overseas institutional investor | 8.78% (28 August 2026 · general investment purpose) |
| Shares issued | 24,396,458 · treasury 20,005 · outstanding 24,376,453 |
| Paid-in capital | KRW 12,198 mn (USD 8.7 mn) · unchanged over the past five years |
Solborn Inc. was founded in 1994 and listed on KOSDAQ in 1999. It discloses its business as management participation through share ownership, plus real estate leasing.
1-1What it makes and what it replaces
What the company makes is software that stores, transmits, and retrieves images digitally. What it replaces is the entire job of developing film, storing it, and carrying it around. Images produced by imaging equipment are received in the international DICOM format, stacked on a server, and called up on clinicians' PCs, tablets, and phones. In the film era you needed space and time to develop images, and finding a prior study meant a trip to the archive room. Now the image appears on screen as it is captured, and several doctors can open the same study at once. The product line widened sideways from this single root.
| Product line | What it replaces |
|---|---|
| PACS by clinical department | Started in radiology, extended to cardiology, pathology, ophthalmology, dentistry, veterinary, and gastroenterology. Each department's reading workflow is handled separately |
| Dose management (DoseM) | Automatically records and manages the radiation dose a patient receives. US state and federal dose-management rules and the EU radiation safety directive are the demand basis |
| Vendor Neutral Archive (VNA) | Gathers images scattered across departments and institutions into one place. The company states in filings that it was the first to launch this in Korea |
| Five AI reading-assist products | First-pass screening for the reading physician. Includes both in-house products and third-party AI |
The company holds Class II medical device clearances from the US Food and Drug Administration (FDA) and the European CE mark, plus approvals from Korea's Ministry of Food and Drug Safety (MFDS) and ISO 13485, 9001, 27001 and 27701. Every product has gone through this process since US exports began in 2000.
1-2Where this product sits inside hospital IT
This software is one layer of hospital IT. Registration, orders, clinical records and billing are handled by the electronic medical record (EMR) and hospital information system (HIS), which are not this company's products. The workflow runs like this — a doctor orders a study in the EMR, that order passes into the study worklist, the images captured by the equipment accumulate in this company's system, and the reading result returns to the EMR as text. Because a single study can run to hundreds of megabytes and cannot sit in the same container as order records, the two systems stand apart and are joined by the HL7 international messaging standard. Replacing this company's system means relaying all of those interfaces as well.
The company is trying to widen out from this position. "Medical information business" is registered as an active purpose in its articles of incorporation, and the company describes its INFINITT Healthcare Platform as one that "integrates and stores medical images and information data scattered within a healthcare institution in line with international standards" (company description in filings). The company states that INFINITT MDT, a multidisciplinary care system, has been supplied to and is running at institutions in Korea, the US and Europe. Its patent list also includes items that handle data beyond images.
Sources · Annual report I-1~4, II-1, VII · intellectual property schedule · semi-annual report I-4, VII · company description in filings
Business model
How it makes money
It sells hospitals software that is hard to replace once installed
What it sells is hospital software built around a Picture Archiving and Communication System (PACS). Around that core sit department-specific PACS, dose management (DoseM), the Vendor Neutral Archive (VNA), and five in-house AI reading-assist products, with implementation, equipment integration and maintenance attached. The company states it has supplied more than 6,300 institutions in 55 countries (company claim · no external source to check it against).
Money flows in this order: initial build (one-off) → maintenance (recurring) → upgrade every three to five years. The company does not disclose the weight of each layer, but revenue recognition splits them. H1 2026 revenue "transferred over time" was KRW 37,320 mn (USD 26.7 mn), 64.8% of sales and up 23.8% year on year. Short-term advances and deferred revenue, which are contract-liability in nature, also rose 16.1% from KRW 22,170 mn to KRW 25,740 mn.
FY2025 revenue mix — services (maintenance and implementation) KRW 60,610 mn (USD 43.3 mn), 59.2% · merchandise (server and equipment resale) KRW 21,150 mn (USD 15.1 mn), 20.7% · products (licences) KRW 19,880 mn (USD 14.2 mn), 19.4% · other KRW 760 mn, 0.7%. The low-margin merchandise share fell to 12.1% in H1 2026, and as a result the gross margin rose from 61.4% to 67.6%. The 4.5% revenue decline in Q2 2026 came from the same line — merchandise revenue −71.9%.
Sources · Annual and semi-annual report note 38
Market and competition
Replacement at home, growth abroad
Korea's PACS market formed when reimbursement was applied under national health insurance in November 19994, and adoption at hospital level was essentially complete by around 2005. In other words this is a replacement and upgrade market, not a new-adoption market.
Revenue by region (KRW mn)
Sources · Annual report note 38 (FY2023 from the amended filing, receipt no. 20260320001108) · semi-annual report note 38 · FY2024 omitted for space (Korea 41,350 / Americas 29,710 / Europe 8,970 / Asia 15,720)
All of the growth came from outside Korea. Three forces could enlarge the field, and the quality of evidence behind each differs. ① Overseas expansion — exports rose from KRW 46,770 mn to 55,790 mn to 62,680 mn (USD 44.8 mn), a three-year compound average of 15.8%, lifting the export share of revenue from 54.5% to 61.2%. This is the only engine measurable from filings, and the largest. ② Departmental expansion — a structure that sells more into the same hospital. The product line-up is a disclosed fact, but the revenue contribution is not disclosed. ③ AI add-ons — a chest CT reading deployment in North America in August 2026 and two digital pathology collaborations, still early, with revenue and adoption rates undisclosed.
Note: the export/domestic split (annual report II-4) is compiled on a different basis from the regional table above (note 38), so the domestic figure differs slightly. The two tables should not be added together.
Ceiling and headwinds — Third-party estimates of the global PACS and RIS (Radiology Information System) market range from USD 2.45 bn to USD 6.58 bn for 2025, a wide spread, with growth rates clustered in the 4–9% a year range. Because each research house defines the market differently (PACS only / PACS+RIS / including VNA), these were not used as a denominator for the company's share, only as a reference coordinate. The ceiling has two layers — at home, what remains is only replacement demand; abroad, the ceiling is higher but much of the market sits in developed countries, where the company meets the global majors head-on.
3-1Competitors by layer
| Layer | Competitor | Confirmed facts | Where this company stands |
|---|---|---|---|
| Domestic large-hospital replacement | Taeyoung Soft (ZeTTA PACS) | Entered 2015 · 7 of 47 tertiary general hospitals in the 5th cycle · replacement across seven hospitals of one Catholic medical centre (Aug 2021) · next-generation project at another medical centre (Jan 2022) · Wonju Severance (Mar 2025) · FDA clearance | Losing ground [press 2019–2025] |
| Domestic small hospitals and public health centres | Medical Standard | Described as number two in Korea · no share figure | Competes in a different segment [press] |
| Domestic overall | About six vendors | 98% penetration at large hospitals — a replacement market, not a new one | Number one position (2019 basis) [press] |
| Global top tier | GE HealthCare · Philips · Fujifilm · Agfa · Carestream | Top five combined roughly 47% | Behind at large global institutions [compiled · wide variance] |
| Overseas quality ratings | Sectra · Agfa and others | Upper ranks at KLAS Research, a US healthcare IT research firm | Rebounded to third in the small-volume segment in 2025 |
Sources · Annual report II competitive landscape (domestic competitors described only as "many small and mid-sized firms," with no names) · press · third-party research compilations
Substitute technology — because the DICOM standard is open, technology itself does not block entry. There is really one substitution path: a cloud platform turning the migration of existing image data into a product that is low-cost and causes no downtime. At that point, replacement drops for a hospital from "a major undertaking" to "a purchase order."
Abroad, the direction is the opposite. The company's KLAS score in the PACS segment (under 300,000 studies a year) rose from 79.3 (6th) in 2024 to 88.8 (3rd) in 2025, and it ranked first in cardiovascular imaging in the 2026 Best in KLAS awards with 89.0. The external quality indicator, empty for six years after the "12-time number one through 2019" line in the periodic report, has been filled in again. That said, the Americas — the largest region by revenue — grew 6.2% in H1 2026, which reads as market speed [Interpretation], and Europe's +47.4% is the real engine of growth.
The moat
How long can this position hold?
| Source of moat | Strength | What is confirmed |
|---|---|---|
| Switching costs | Strongest | Runs 24/7/365 · if it stops, reading stops · tied by interfaces to EMR and HIS · replacement requires migrating petabytes of data again |
| Economies of scale | Strong | With revenue up 19.2% over three years (KRW 85,860 → 102,390 mn), operating profit rose +261% (KRW 3,940 → 14,220 mn) · gross margin 67.6% · raw material purchases 5.8% of sales |
| Regulation and certification | Present | FDA and CE Class II medical device · four ISO certifications · gates that take time to clear have already been cleared |
| Technology gap | Present but thinning | Competition happens on top of an open standard, so the standard itself is not a barrier · R&D spend fell three years running (KRW 9,930 → 9,830 → 7,600 mn) |
| Brand and reputation | Double-edged | KLAS number one 12 times (2009–2019), then a six-year gap → third in 2025 · first in cardiovascular imaging in 2026 / disclosure on 17 October 2025 of alleged embezzlement and breach of trust (annual report XI-3 · receipt no. 20260320001108 — this is at the allegation stage and is not a settled judicial finding), after which reputation was in fact damaged |
| Network effects | Not counted | Image sharing between institutions is in operation, but the number of connected institutions and traffic are undisclosed. Not counted as a moat |
Overall — the current level is on the strong side, but the two pillars that built it are thinning at the same time. ⓐ The investment that creates the technology gap has fallen for three years, and ⓑ reputation has been through an allegation and a trading suspension. There is a fact on the other side too — switching costs do not shrink no matter what the company does. What it costs a hospital to change systems has nothing to do with this company's R&D budget, and institutions already installed will not disappear within five years. The author reads this as close to strong today and average in three to five years [Interpretation].
Environmental tailwinds point the right way, but their size is unconfirmed. Wider AI reading and demand for image integration only turn into money on top of the installed base, and that contribution is entirely undisclosed. Dose-management regulation is now in its tenth year, so it is not a new change. Neither axis is zero, but there is still not a single measured instance of a tailwind reaching volume or price.
Two ways the moat weakens — ⓐ If any vendor turns data migration into a low-cost, no-downtime product, what remains is price competition, and this company described its price trend as "not reasonably determinable," which means there is no standard price list. ⓑ If EMR and HIS vendors extend image viewing. Taking over "reading," where screens and tools differ by department, is hard, but even bolting on "viewing" puts pressure on the price this company's product commands. Whether this is actually happening was not verified this time (see 9-1).
Financials
The current state of health in numbers
5-1One profit line with two natures mixed into it
The income statement reads as if two companies were laid on top of each other. The upper half (operating) is a software business that generates cash. The lower half (non-operating) is a portfolio of overseas ETFs.
Upper half — 96.3% cash conversion. Operating cash flow rose three years running, from KRW 10,220 mn to 11,810 mn to 13,690 mn (USD 9.8 mn), and jumped from KRW 890 mn to 8,630 mn in H1 2026. Capital expenditure (CAPEX) is only KRW 360–720 mn a year, so almost all of it survives as free cash flow. On an FY2025 basis, operating profit converts to operating cash flow at 96.3%, and operating cash flow to free cash flow at 96.8%. The profit the core business makes is cash.
Lower half — KRW 95,130 mn unrealised. Financial assets at fair value through profit or loss (overseas ETFs and similar) carry an acquisition cost of KRW 102,590 mn (USD 73.3 mn) and a fair value of KRW 197,720 mn (USD 141.2 mn). The difference of KRW 95,130 mn (USD 68.0 mn) has never once become cash. During H1 2026 the company bought KRW 9,540 mn and sold KRW 110 mn, so the balance grew. Five-year (FY2021–2025) cumulative free cash flow of KRW 57,490 mn ÷ cumulative net income attributable to owners of KRW 103,460 mn = 55.6%. About half of the profit is an accounting valuation.
5-2Dissecting the SG&A decline — a one-off, or a return to earlier levels?
| SG&A item (KRW mn) | H1 2026 | H1 2025 | FY2024 | How to read it |
|---|---|---|---|---|
| Bad debt expense | 200 $0.1m | 1,820 $1.3m | 150 $0.1m | H1 2025 was the outlier · H1 2026 is back at the 2024 level |
| Fees paid | 2,730 $1.9m | 4,370 $3.1m | 6,110 $4.4m | Near the 2024 half-year level (3,060) |
| Retirement benefits | 510 $0.4m | 470 $0.3m | 1,120 $0.8m | The −1,360 in Q2 reverses the +1,870 in Q1 · net half-year effect zero |
| Salaries | 10,410 $7.4m | 10,400 $7.4m | 21,440 $15.3m | Flat · headcount 246 → 242 |
| Total | 24,570 $17.6m | 26,530 $18.9m | 51,010 $36.4m | SG&A ratio 42.6% = the 2021–22 level (40.6–43.7%) |
Sources · Semi-annual and annual report note 28 — [Interpretation] because both lines returned to earlier levels, this reads as costs going back to where they were. There is room to read it the other way.
5-3Capital structure, shareholder returns, potential dilution
Borrowings of KRW 570 mn against cash and short-term financial instruments of KRW 19,930 mn leave net cash of KRW 19,360 mn (USD 13.8 mn), with a debt-to-equity ratio of 27.79%. Annual free cash flow of +KRW 13,260 mn means no cash burn.
On shareholder returns, there was no year-end dividend for FY2023 or FY2024; FY2025 began with a quarterly dividend of KRW 2,440 mn plus a year-end dividend of KRW 2,440 mn (together USD 3.5 mn). The record date for the 2026 quarterly dividend is 23 October 2026, with the amount undecided. For the 20,005 treasury shares, the filing explicitly states there is "no plan to purchase or cancel."
Dilution is effectively absent. There are no convertible bonds, bonds with warrants, or redeemable convertible preferred shares, and only 90,000 stock options (0.37%) remain outstanding.
5-4By entity — growth came from overseas subsidiaries, not the parent alone
The company reports a single segment (software), so there is no segment-level profit disclosure. This section therefore splits the business only along the three axes that disclosure actually provides — standalone versus consolidated, by subsidiary, and by timing of revenue recognition. No divisional breakdown was constructed.
① Standalone versus consolidated — standalone revenue has sat near KRW 50,000 mn for three years
| KRW mn | FY2024 | FY2025 | H1 2026 | How to read it |
|---|---|---|---|---|
| Consolidated revenue | 95,750 $68.4m | 102,390 $73.1m | 57,630 $41.2m | +19.2% over three years |
| Standalone (parent only) revenue | 51,070 $36.5m | 50,370 $36.0m | 25,160 $18.0m | Flat (annualised 50,330) |
| Difference (subsidiaries and consolidation adjustments) | 44,680 $31.9m | 52,020 $37.2m | 32,470 $23.2m | All of the growth is here |
| Standalone net income | 38,240 $27.3m | 27,490 $19.6m | 29,540 $21.1m | Of standalone assets of 250,130, ETFs are 197,060 — the ETFs sit in the parent's accounts |
Sources · Semi-annual report II-1 summary financial information
② By subsidiary — where it sells and where it keeps the money are different places
| Entity (H1 2026 · KRW mn) | Revenue | H1 net P/L | Equity | Note |
|---|---|---|---|---|
| North America | 21,990 $15.7m | +3,100 +$2.2m | −2,400 −$1.7m | Largest overseas entity · FY2025 full year was a net loss of −780 · capital deficit |
| Europe | 6,750 $4.8m | +1,430 +$1.0m | +1,590 +$1.1m | Engine of H1 2026 growth (+47.4%) and a 21% margin |
| Domestic subsidiary (Techheim) | 3,420 $2.4m | +870 +$0.6m | +10,070 +$7.2m | Thickest equity among subsidiaries |
| Japan | 2,310 $1.6m | −100 −$0.1m | −1,980 −$1.4m | FY2025 also −160 · two straight loss-making years |
| Southeast Asia | 2,080 $1.5m | −630 −$0.5m | +670 +$0.5m | Swung to a loss from +570 in FY2025 |
| Taiwan | 1,220 $0.9m | −80 −$0.1m | −110 −$0.1m | Small |
| Middle East & Africa | 1,120 $0.8m | +20 +$0.0m | −5,290 −$3.8m | Capital deficit is 4.7× revenue |
| Shanghai | 0 | 0 | −8,130 −$5.8m | Assets 0 · liabilities 8,130 (prior year 7,340) · creditor undisclosed · 0.3% of total assets |
Sources · Semi-annual report notes 1 and 11-12 — 11 subsidiaries (8 overseas, 3 domestic), all unlisted
The subsidiaries' H1 2026 net results sum to roughly +KRW 5,300 mn. The gap against consolidated net income of KRW 38,740 mn comes mostly from ETF valuation gains in the parent's accounts. In other words, overseas entities generate the revenue, while financial assets in the parent's accounts set the size of the profit [Interpretation].
③ By timing of revenue recognition — revenue transferred over time (recurring) was KRW 37,320 mn, up 23.8%, lifting its share from 58.9% to 64.8%. Note that this line contains both maintenance and percentage-of-completion recognition on multi-year implementation work, so pure maintenance cannot be isolated. Revenue transferred at a point in time (one-off) was KRW 20,310 mn, down 3.5%; looking at Q2 alone it fell 37.8%, from KRW 11,870 mn to 7,390 mn. The decline is concentrated in the hardware resale line, while the software core did not break (products +4.3%, services +22.3%).
Sources · Semi-annual report note 38 (H1 2026 cumulative versus H1 2025 cumulative)
5-5Revenue forecast
What follows is the author's own estimate, not company guidance. The company does not publish revenue or profit targets in its periodic reports. The three scenarios differ only in three lines — revenue, cost ratio and SG&A ratio — while net financial income, equity-method results, tax rate and minority interest ratio are shared. ETF valuation gains and losses are set to zero from 2027 onward, because they are market-price movements whose direction the company does not control.
FY2028 end point (KRW mn · author's own estimate)
| Case | Revenue | Operating profit | Net income attrib. | Assumption in one line |
|---|---|---|---|---|
| Bull | 138,510 $98.9m | 34,590 $24.7m | 29,870 $21.3m | Exports hold the measured three-year average of 15.8% + H1 2026 margin structure holds (cost ratio 32.4%, SG&A ratio 42.6%) |
| Base | 127,620 $91.2m | 27,670 $19.8m | 24,320 $17.4m | Regional trends carry over (Korea −1.3%, Americas +12.9%, Europe +13.6%, Asia +6.4%) · costs move with revenue |
| Reference · no growth | 108,820 $77.7m | 23,700 $16.9m | 21,140 $15.1m | A conservative alternative reflecting only confirmed measured increments and flat-lining after 2026 |
| Bear | 105,840 $75.6m | 18,190 $13.0m | 16,730 $11.9m | Merchandise revenue sets at the H1 2026 level and the SG&A ratio returns to the FY2025 actual of 47.61% (two lawsuits in progress) |
Base case by year (KRW mn · author's own estimate)
| KRW mn | FY2026E | FY2027E | FY2028E |
|---|---|---|---|
| Revenue | 111,890 $79.9m | 119,260 $85.2m | 127,620 $91.2m |
| Operating profit | 24,260 $17.3m | 25,860 $18.5m | 27,670 $19.8m |
| Operating margin | 21.7% | 21.7% | 21.7% |
| Net income attributable | 45,750 $32.7m | 22,870 $16.3m | 24,320 $17.4m |
| Free cash flow | 22,860 $16.3m | 24,400 $17.4m | 26,140 $18.7m |
FY2026 net income attributable is unusually large because ETF valuation gains of KRW 27,990 mn (USD 20.0 mn), already fixed in the first half, are common to all three scenarios. The 2027 drop (45,750 → 22,870) is not a picture of the business deteriorating; it is a step created by setting valuation gains to zero. Separate stress case — if the ETFs repeated a valuation loss of the size measured in FY2022 (−KRW 11,270 mn), base-case net income attributable would fall to roughly KRW 12,100 mn (USD 8.6 mn).
Three possible futures
Why the upside and downside are not symmetric
If the bull case plays out — if exports hold their measured three-year pace and AI modules start showing up in revenue, the company moves from "Korea's number one PACS vendor" to "a company that sells modules on top of an installed base." More money comes out of the same hospital.
If the base case plays out — Korea stays flat while overseas entities grow gently. Revenue rises around 7% a year and the margin structure holds. The character of the company is the same as today; only the size changes.
If the bear case plays out — software fails to fill the space left by hardware resale, dispute costs come back into SG&A, revenue flat-lines and profit returns to 2025 levels. What is left then is not a business but an entity holding assets.
Why the range is skewed — measured on final-year net income attributable, the upside is +23% and the downside −31%. The downside is wider because the pillars are made of different material. The two premises holding up the bull case (exports at 15.8%, AI modules turning into revenue) have never once been confirmed, whereas the two grounds under the bear case (falling merchandise revenue, ongoing dispute costs) are extensions of things that have already happened. An unconfirmed assumption and an event that has already occurred do not carry the same weight.
| Bull-side facts | Bear-side facts |
|---|---|
| Recurring revenue KRW 37,320 mn (64.8% of sales) · +23.8% · contract liability balance +16.1% | Five-year cumulative FCF ÷ net income attributable 55.6% — half of profit is an unrealised valuation |
| Gross margin 61.4% → 67.6% — driven by shrinking low-margin hardware | Standalone revenue flat at 51,070 → 50,370 · consolidated growth belongs to overseas subsidiaries |
| Operating profit 3,940 → 14,220 over three years · 96.3% cash conversion | H1 2026 ETF purchases 9,540 versus disposals 110, plus 1,010 of additional parent-company shares — no measured instance yet of capital returning to the core business or to shareholders |
| Net cash 19,360 · debt-to-equity 27.79% · effectively no dilutive securities | Long-term trade and other non-current receivables 5,750 → 11,520 (+100%) · allowance ratio 14.6% → 22.6% |
| Europe +47.4% · KLAS 79.3 → 88.8 · first in cardiovascular imaging in 2026 | Shanghai revenue 0 · liabilities 8,130 · equity −8,130 · Japan, Taiwan and Southeast Asia loss-making in H1 2026 |
| Dividend initiated in FY2025 (4,880) · credit rating up two years running | Control-related litigation in progress (appeal 2026Na204361 · petition for court permission to convene an extraordinary general meeting, 2026Bihap2155) · six claims against the company totalling about 3,650 |
| Auditor's key audit matter for three consecutive years: "appropriateness of disclosure of related-party transactions and balances" · H1 2026 related-party purchases 1,090 (of which 710 from the parent) | |
| Disclosure on 17 October 2025 of alleged embezzlement and breach of trust → trading suspension · listing eligibility review5 → listing maintained and trading resumed on 20 January 2026 · designated an unfaithful disclosure corporation6 on 8 December 2025 (penalty KRW 32 mn). This is at the allegation stage and is not a settled judicial finding. |
What to watch
Which signals, and when — the answer is tied to the November quarterly report and to the courts
| Variable | Base assumption | Fork in the road | Decision point |
|---|---|---|---|
| Q3 export and domestic revenue | Second half = prior-year level | Double digits keeps the bull axis alive / two consecutive quarters of export decline points to the bear path | Q3 2026 report, November |
| Direction of merchandise versus products and services | Mix shift continues | If only merchandise falls while products and services hold, the "mix shift" reading firms up / if all three fall, it is demand slowing | Q3 2026 report, November |
| ETF valuation gains and losses | Zero from 2027 | Fair value 5% = ±9,890 · FX 10% = ±21,000 (company-disclosed sensitivity) | Quarterly filings |
| Court permission for an EGM · treasury share purchase agenda | Zero realised | Permission opens a return channel / rejection weakens the pressure | Ad hoc · court decision |
| Selection of vendors for tertiary hospital replacements | Domestic share stagnant | Further losses mean lower domestic volume — the domestic lever in the bear scenario lights up | Ad hoc · filings and press |
| Appeal judgment | Dispute continues | Depending on the outcome, dispute costs go either way | Seoul High Court 2026Na204361 |
| North American entity's capital deficit | Deficit persists | Escaping it as Europe did aligns where growth is with where profit is / staying in deficit keeps them apart | Annual report, March 2027 |
| KLAS ranking | Holds at the 2025 level | Holding supports the view that lower R&D did not damage the product / slipping back says the opposite | Next release |
| Separate disclosure of AI and VNA revenue | Zero | If it begins, the bull case's second premise becomes verifiable for the first time | Annual report, March 2027 |
The bottleneck, in one line — the heaviest item is the first row (the direction of Q3 revenue). The size of the other eight is only set once that row resolves.
Summing up
The market's received view versus what the author verified
On one page — the confirmed strengths are switching costs, recurring revenue of KRW 37,320 mn (64.8%), a 67.6% gross margin, 96.3% cash conversion, net cash of KRW 19,360 mn, a 27.79% debt-to-equity ratio, and a credit rating raised two years running. The unverified weaknesses are that the maintenance renewal rate, volumes, prices, the current domestic share, and AI and platform revenue are all undisclosed, plus three years of falling R&D, three years of flat standalone revenue, and a five-year FCF-to-net-income ratio of 55.6%. Opportunities are departmental expansion, the integrated platform direction, profit realisation at the European entity, dose-management regulation, the rebound in quality ratings, and the resumed dividend. Threats are Taeyoung Soft's encroachment at home, growth in the Americas at market speed, possible extension of image features by EMR vendors, ETF swings, control-related litigation, and four overseas entities in capital deficit. That is 14 good boxes against 13 bad ones — and the distinctive thing is that the strengths are mostly confirmed in numbers while the weaknesses are mostly of the form "cannot be confirmed."
The market's story is generally threefold: ① the core business is fine but the money earned goes into overseas ETFs ② the holding company and founder use the listed subsidiary as a private treasury ③ activism has arrived and things may change.
① is true, except that it leaves out the part where the core business has not merely stayed "fine" — first-half gross profit rose KRW 7,530 mn and the SG&A ratio returned to 42.6%, the 2021–22 level. On ②, what disclosure confirms is three things: related-party purchases of KRW 1,090 mn (of which KRW 710 mn from the parent), an additional KRW 1,010 mn of parent-company shares acquired, and the same key audit matter for three consecutive years. "Private treasury" is an interpretation of those facts, not a disclosed fact. On ③, the contest for shares is real — the largest-shareholder side holds 51.80% and an overseas institutional investor 8.78%, both having bought, and a petition for court permission to convene an extraordinary general meeting including a treasury share purchase agenda is before the court. Even so, the only money that actually reached shareholders is the FY2025 dividend of KRW 4,880 mn (USD 3.5 mn).
On direction, the author sees revenue rising gently, around +7% a year. Certainty is low, because the regional growth rates are not observed values but the midpoint of a conservative trend and three years of measured data. On verification, recurring revenue and the cost structure have reached the point where they can be checked in numbers, while AI module revenue and the current domestic share still have no channel open.
What would change this picture — ⓐ if Q3 exports fall for a second consecutive quarter, the growth axis itself is shaken ⓑ if Q3 SG&A jumps again, the reading that "costs went back to their earlier level" collapses ⓒ if the court takes up the treasury share agenda, the direction of capital allocation changes for the first time ⓓ if more tertiary hospital replacements are lost, domestic volume falls.
For as long as this remains a software company's income statement whose net income belongs to an overseas index, reading this company stays split in two — reading the business, and reading the capital allocation.
Limitations
Please read this alongside the rest
9-1What could not be verified
The items below are not things the author failed to look for; they are things that could not be found in public sources. This list is the confidence interval of this report.
Maintenance contract renewal rate — the company states only that it is "far higher than in other industries" and discloses no figure. It is the key metric of this business, and it cannot be verified.
Volumes and prices — the price trend disclosure was omitted as "not reasonably determinable." Growth cannot be split into volume and price, so the author's estimates are effectively ratio-based estimates.
Order backlog and major customers — omitted as commercially sensitive information. There is no external indicator that would flag the second half in advance.
Current domestic market share — only the 2019 figure exists. The upper bound can be calculated; the lower bound is unknown.
Revenue contribution of the AI family and the integrated platform — revenue is disclosed in only three categories (products, merchandise, services), so there is no line in which they would appear.
The nature of the Shanghai entity's KRW 8,130 mn of liabilities — the creditor is unidentified. The size is 0.3% of total assets.
The region and customers behind rising long-term receivables — the ageing analysis shows the two-to-five-year bucket actually shrank (7,340 → 6,670), but which market drove the increase is not disclosed.
Global market size — third-party estimates for 2025 differ by a factor of 2.7, from USD 2.45 bn to USD 6.58 bn. Because the definitions differ, they were not used as a denominator.
Whether EMR vendors are extending image features — whether such a trend actually exists was not verified this time against institutional research or filings. Item ⓑ in section 4 records a direction only; it is not a verified fact.
Export controls and tariff exposure — the view that direct exposure is low because revenue comes from software and services is an interpretation, not something verified against institutional sources.
9-2The four places this analysis is most likely to be wrong
㉠ How costs were read — the judgement that costs "went back to their earlier level" rests on a single half-year observation and could be overturned by Q3 SG&A. If this was in fact a temporary decline during a quiet phase of the dispute, then the new level of operating profit is a peak rather than a baseline, and all three paths in section 6 shift downward as a block. This is the place this report is most likely to be wrong.
㉡ The possibility of being wrong in the other direction — second-half cost of sales and SG&A were carried over from the prior year's actuals, and because second-half 2025 SG&A was unusually low, that carry-over already leans in favour of profit. The base case therefore stands slightly more optimistic than it looks.
㉢ Where growth is versus where capital is — the overseas entities where growth occurred and those in capital deficit are largely the same places. Both facts appear in the table in 5-4, but how far the two offset each other was not measured.
㉣ How the inputs were selected, and the axes left untouched — market share was updated downward (2019 basis plus replacement wins) while quality ratings and the credit rating were updated upward, and the two directions rest on different kinds of evidence: the downward side is mainly press reporting, the upward side third-party assessment and filings. In addition, institutional axes such as export controls and tariffs, and movements on the EMR side, were not checked against institutional sources. The last two items in 9-1 are closer to "not touched this time" than to "could not be found."
9-3Verification steps omitted, and why
The valuation steps that produce a number were not performed. This is a report that does not carry a number.
9-4Method and the author's admissions
The primary sources are original periodic and ad hoc filings from DART7, Korea's electronic disclosure system. Press reports are marked [press], and items resting on press reporting alone were not reflected in the estimates.
The five years of measured data used as the basis for estimates are five annual years (2021–2025) plus half-year actuals. The 2021 and 2022 figures are noted as coming from the original pre-restatement filings. Because volumes and prices are not disclosed, ratio substitutes were used in three lines of the estimation procedure.
The base case's regional growth rates are the midpoint of a conservative trend and the three-year measured average. The author notes that these are not observed values but a midpoint between two methods, which is why the no-growth alternative (KRW 108,820 mn) was left standing alongside rather than deleted. Second-half cost of sales and SG&A were carried over from prior-year actuals, and that carry-over leans in favour of profit.
Because there is a single reporting segment (software), the build was split along regional (entity) lines rather than divisions. Market share, the 6,300-institution figure and the renewal rate all rest solely on the company's own statements, with no external channel available to check them.
Key numbers on one page
| Item | Value | As of |
|---|---|---|
| Shares issued · shares outstanding | 24,396,458 · 24,376,453 | 30 Jun 2026 |
| Treasury shares | 20,005 (0.08%) · "no plan to purchase or cancel" stated | 30 Jun 2026 |
| Paid-in capital | KRW 12,198 mn (USD 8.7 mn) · unchanged over five years | 30 Jun 2026 |
| Potential dilution | 90,000 stock options (0.37%) · no convertible bonds, bonds with warrants, or redeemable convertible preferred shares | 30 Jun 2026 |
| Largest-shareholder side | 51.80% | 10 Jul 2026 |
| Net income attributable (5-year cumulative) | KRW 103,460 mn (USD 73.9 mn) | FY2021–25 |
| Order backlog | Omitted from disclosure as commercially sensitive | — |
| Revenue forecast (author's own estimate) | FY2026 111,890 / FY2027 119,260 / FY2028 127,620 (base case) | — |
Share price and market capitalisation are not included. Neither is anything derived by dividing by the share count.
Notes on Korean institutional terms
1 · KOSDAQ — the Korea Securities Dealers Automated Quotations market, operated by the Korea Exchange (KRX). It sits alongside the main KOSPI board and lists mainly technology and mid-sized companies, comparable in role to the Nasdaq.
2 · Tertiary general hospital (상급종합병원) — a status designated by Korea's Ministry of Health and Welfare in three-year cycles for large hospitals that treat severe conditions. The 5th cycle covers 47 hospitals nationwide. Designation carries higher insurance reimbursement, so these hospitals are the largest and most demanding IT customers in the country.
3 · HIRA (Health Insurance Review and Assessment Service, 건강보험심사평가원) — the public body that reviews medical claims and assesses care quality under Korea's national health insurance. Its register of hospitals is the standard denominator for counting institutions.
4 · Reimbursement applied in November 1999 (PACS 수가) — from that date, national health insurance paid hospitals a fee for reading images through a digital system. This effectively subsidised PACS adoption and is why the Korean market filled up early, by around 2005.
5 · Listing eligibility review (상장적격성 실질심사) — a Korea Exchange procedure triggered by events such as an embezzlement or breach-of-trust disclosure. Trading is suspended while the exchange decides whether the company should remain listed. In this case the exchange decided to maintain the listing and trading resumed on 20 January 2026.
6 · Unfaithful disclosure corporation (불성실공시법인) — a Korea Exchange designation for a listed company that fails to disclose, delays, or changes a disclosure. It carries penalty points and, as here, a monetary penalty.
7 · DART (Data Analysis, Retrieval and Transfer System) — the electronic disclosure system operated by Korea's Financial Supervisory Service, where all regulatory filings are published. Equivalent in function to EDGAR in the United States.
Sources
Primary · Financial Supervisory Service electronic disclosure (DART)
- Semi-annual report (25th fiscal year, half year · 14 Aug 2026) — receipt no. 20260814002622
- Annual report (24th fiscal year) and amendment (20 Mar 2026) — receipt no. 20260320001108
- Annual report (21st fiscal year · FY2022 · pre-restatement) — receipt no. 20230323001186
- Petition for court permission to convene an extraordinary general meeting (2026Bihap2155 · 31 Aug 2026) — receipt no. 20260831900766
- Litigation judgment/decision filing (appeal 2026Na204361 · 31 Aug 2026) — receipt no. 20260831900772
- Determination of record date for interim (quarterly) dividend (21 Aug 2026) — receipt no. 20260821900735
- Large-holding report · largest-shareholder side 51.80% (20 Jul 2026) — receipt no. 20260720000308
- Large-holding report (abbreviated) · overseas institutional investor 8.78% (1 Sep 2026) — receipt no. 20260901000108
Secondary · press and third-party material (not reflected in estimates)
- KLAS PACS segment score comparison (2024 versus 2025 · Small 79.3 → 88.8) — intuitionlabs.ai, Radiology Business
- 2026 Best in KLAS · cardiovascular imaging 89.0 — KLAS Research
- Korean tertiary hospital PACS replacement trends (2019 onward) — Medical Times, Rapportian, Electronic Times
- Global PACS/RIS and VNA market size estimates — Precedence (USD 2.45 bn), Spherical (USD 3.82 bn), market.us (USD 6.58 bn), VNA+PACS (USD 4.44 bn)
- Reporting on changes in the largest-shareholder side's stake — Digital Today
Market size estimates differ widely across research houses in both definition and value, so they are carried only as a reference coordinate. None was averaged or singled out.
External Business Analysis Report · INFINITT Healthcare Co., Ltd. · As of 1 September 2026 · No fair value or target price is calculated
댓글
댓글 쓰기