Imagine a tollbooth beside a city gate. Everything that enters the city passes through it, and the booth’s owner collects the toll. One day the lord decrees that his close kin may no longer own tollbooths. The booths go on sale, merchants crowd in, and the largest one is bought by the lord’s nephew, a relative standing just outside the decree’s reach. The merchant who overpaid for the next booth learns late that the lord still holds the key to the gate, and can open another one whenever he likes.
In financial language: the tollbooth is a Korean hospital GPO (gannapsa, a purchasing intermediary that sits between a hospital and its medical-supply vendors), the lord is the hospital, and the decree is Korea’s amended Medical Device Act. What the merchant bought is 51% of a company. What he actually holds is a single supply contract the hospital can take back. Earlier this year the desk argued, in a sector deep-dive, that this market was not a roll-up of assets but a roll-up of contracts, and that the largest volumes would be allocated by families rather than by markets. This autumn two transactions tested that reading.
This article reads EZmedicom’s reported pursuit of Dongha Medical and Hyundai Green Food’s selection as preferred bidder for MediGood Partners, not for their headline valuations but for the control mechanics underneath. The question is not what 51% costs. It is who keeps the other 49%, and what that remainder can still do.
The Deal
Three transactions, one regulation. The amended Medical Device Act, effective December 2027, bars hospitals from procuring through intermediaries that are their special related parties. Hospital foundations are therefore selling majority stakes in their captive supply arms. The three cases below are the same event at three different settings.
- Dongha Medical (Soonchunhyang). A supplier of clinical materials to Soonchunhyang hospitals since 1994, spun out of Dongha Industry’s medical materials division at the end of 2025. Dongha Industry is 100% owned by the Dongeun Foundation, the school foundation behind Soonchunhyang. Last year it supplied KRW 24 billion (approx. USD 17 million at KRW 1,400/USD, October 2026) to Soonchunhyang University Bucheon Hospital alone, 99.57% of that hospital’s purchases. On offer: 51% plus control. The market estimates enterprise value at around KRW 70 billion; the seller reportedly asks at least 10x on annual operating profit of roughly KRW 7 billion, and some see the price approaching KRW 100 billion. The field has narrowed to four bidders: two private equity funds, one diagnostic-healthcare company, one hospital-GPO specialist. EZmedicom is reported to be pursuing the deal; the final buyer is not confirmed.
- MediGood Partners (Seoul Asan). A subsidiary newly formed by the Asan Social Welfare Foundation in October 2025. The foundation subscribed KRW 6 billion of redeemable convertible preferred shares (RCPS) in March; total paid-in capital is disclosed at KRW 6.5 billion. A 51% stake in that entity has reportedly been priced above KRW 100 billion, and Hyundai Green Food was selected preferred bidder over EZmedicom and Serveone. The company supplies more than 2,000 clinical items to Seoul Asan Medical Center, Korea’s largest hospital.
- Avison Care (Yonsei). Acquired by EZmedicom this year. Yonsei University, which had founded the company, sold its entire holding of 51%; the remaining 49% was already held by a third party whose identity has not been disclosed. In 2024 the company posted operating profit of KRW 16.8 billion, donations of KRW 13 billion, and net income of KRW 0.4 billion. Reported prices conflict: KRW 100 to 150 billion in some reports, KRW 250 billion in others.
- The buyer’s balance sheet. EZmedicom reported 2025 consolidated revenue of KRW 501.1 billion, operating profit of KRW 18.8 billion and equity of KRW 87.5 billion; its largest shareholder is, per press reports, on the Daewoong Pharmaceutical side. It acquired a three-month-old entity for KRW 25.8 billion and later explained it as the special-purpose vehicle for the Avison acquisition, with the balance to be funded by financing or other investment whose details are confidential.
Two facts about the rule frame everything that follows. First, the amended Act does not stop at a 50% ownership line; dominant influence over board composition or business operations also constitutes a special relationship, and lawyers interpret that cutting a stake below 50% alone may not fully resolve a violation. Second, the kinship line is drawn at the second degree. The Asan Foundation is a public-interest corporation inside the HD Hyundai group under Fair Trade Commission designation; Hyundai Green Food belongs to the Hyundai Department Store Group; the relationship between the two groups’ controlling persons is uncle and nephew, the third degree. The two are separate business groups under the Fair Trade Act. The MediGood transaction is, on those terms, legally clean.
Structure Breakdown
The centre of gravity in Dongha is not the KRW 7 billion of operating profit. It is that after 51% is sold, 49% stays with Dongha Industry, whose owner is the foundation that established the hospital. The hospital is not selling its stake; the hospital’s foundation is staying at 49% and choosing who gets the other 51%. Reports that Soonchunhyang Medical Center itself grants due diligence to the shortlist and selects the preferred bidder through a final round are a natural consequence of that structure. That naturalness is the problem: a foundation that holds 49% and runs the sale process is leaving a record that looks like dominant influence, and a buyer of 51% could still be found to be a prohibited counterparty. This is an interpretation, but the process itself is writing the evidence for it.
The character of the profit compounds the problem. Even among the bidders there is concern that KRW 7 billion is close to a peak that owner-management built over decades of personal effort. That profit was produced not by a corporate system but by a relationship between specific people and a specific hospital. Shares transfer; relationships stay attached to people. To an architect, the price tag reads not as 10x but as 10x on a non-transferable asset. And with four bidders competing, the price has already begun to reflect the value of a deadline rather than the value of a business: the December 2027 deadline manufactures the buyer’s impatience before it manufactures the seller’s desperation.
MediGood is the mirror image. A company capitalised at KRW 6.5 billion has had its 51% priced above KRW 100 billion. What was sold was not a company. It was the contract to supply more than 2,000 items to the country’s largest hospital. The foundation remains at 49%, and the contract has travelled one full circle inside the family and come back to where it started. The foundation’s statement that the process was fair and transparent, and the industry’s description of it as a closed league, are two faces of the same fact.
Avison Care is the third setting, and it differs from both. Yonsei did not stay at 49%; it exited entirely, and the remaining 49% belongs to an undisclosed third party. Of KRW 16.8 billion in 2024 operating profit, KRW 13 billion left the company as donations, leaving KRW 0.4 billion of net income. The recipient is not disclosed, but the fact stands: nearly eight tenths of operating profit exited under the name of donations. A hospital GPO’s profit is not its accounting operating profit; it is the profit before leakage, and who controlled the leakage channel was the entire price. Whether EZmedicom paid KRW 100 billion or KRW 250 billion, a large part of that sum prepaid the expectation that the leakage would stop, or would accrue to the new owner.
Set the three 49 percents side by side and the real order of this market appears. MediGood: the foundation stays and a buyer within the kinship range takes 51%, so contract-termination risk is lowest and reputational risk is highest. Dongha: the foundation stays but the buyer is an outsider, so dominant-influence risk and peak-profit risk arrive together. Avison: the hospital side has left, so regulatory risk is lowest and commercial risk is highest, because there is no hostage to keep the contract alive. EZmedicom, an incumbent supplier to Seoul Asan, lost the MediGood auction not for lack of capital or track record but because that table was never one that price decides. What remains for it is the hospitals without a family, Soonchunhyang, Eulji, Ajou, where it competes with Serveone and private equity and asks are drifting from the KRW 50 billion range toward KRW 100 billion.
Finally, the weight of capital holding this up. A company with KRW 18.8 billion of operating profit and KRW 87.5 billion of equity has paid, per reports, KRW 100 to 250 billion for Avison and is prepared to pay up to KRW 100 billion for Dongha. That is a picture of up to KRW 350 billion of contract assets stacked on borrowing, and the collateral is a relationship with a hospital that either still holds 49% or has already walked away. Not the shares but the contract; not the contract but the character of the 49%; and beneath both, the maturity of the debt that carries them.
The Governing Logic
An architect structures a deal not to earn but to avoid losing, and reading the capital allocation in these transactions backwards shows what each party was trying to protect. This is a downside-hedge structure rather than an exit narrative, for one reason: what is bought and sold here is not an asset that produces income but a contract that can vanish. For an asset, you design growth. For a contract, you first design against extinction. And because the 49% comes in three characters, the device that locks the downside must come in three forms.
Downside Protection
The MediGood form: foundation retained, buyer within the family. Hyundai Green Food’s financial room is not in doubt: with 2025 consolidated operating profit of KRW 106.8 billion and a net cash position, an acquisition in the KRW 100 billion range is about one year of operating profit, absorbed without borrowing. But the desk’s inference is that the real downside defence sits not in the contract but in the bloodline, the expectation that Seoul Asan will not lightly cut a contract with a company in the extended Hyundai family. On the seller’s side, the foundation chose RCPS over common equity in March; a preferred instrument with a redemption right converts if things go well and recovers principal if they do not, so if the foundation still holds it after the sale it has a second exit on top of its 49%. The sale terms are undisclosed, so whether that preferred functions as a hedge remains interpretation. What is clear is that the foundation keeps 49% and the right to select its vendor, and the buyer has paid more than KRW 100 billion against a relationship the law does not bind. A relationship the law does not bind is a relationship the law does not protect; the moment family dynamics shift, or the closed-league criticism reaches the foundation’s tender transparency itself, that collateral becomes an unenforceable promise.
The Dongha form: foundation retained, buyer an outsider. Here there is a hostage. The Dongeun Foundation holds 49% through Dongha Industry, so obligations can be attached to that 49%. If the desk were entering this acquisition as a co-investor, the design would run four layers deep. First, decompose the consideration: instead of paying the KRW 70 to 100 billion ask outright, split it into fixed consideration of roughly KRW 23 billion for 51% and a deferred KRW 10 billion conditional on three years of maintained supply contracts with the three Soonchunhyang hospitals and a margin floor. If the seller will not underwrite contract continuity, the deferred amount is never paid. Second, a put option on the 49%: if the foundation side terminates the contract, converts to a rebid, or diverts three tenths or more of volume elsewhere, the shareholders’ agreement obliges the 49% holder to buy back the 51% at invested capital plus a cumulative return. The moment the foundation must repurchase the shares if it cuts the contract, tail risk migrates from the buyer to the hospital. Third, a firewall against dominant influence: exclude the foundation’s director-nomination rights, seat an independent chief executive, and have trading terms verified externally, so that any record of the 49% exercising real power is structurally erased. This is regulatory defence and a device that makes the buyer’s control real at the same time. Fourth, a put triggered by regulation itself: if legislation banning the 49% holding clears committee, or a margin-cap notice is issued, the investor must be able to hand its stake to the operator at principal plus a defined return and leave.
The Avison form: hospital side gone. There is no captive here. Yonsei sold all of its 51% and left, and the owner of the remaining 49% is undisclosed. With no counterparty against which to write a put on the 49%, the downside can only come from the contract document itself: the remaining term of the long-term supply agreement, minimum volume guarantees, compensation on early termination, and whether the KRW 13 billion of donations continues to flow out under the new owner. EZmedicom’s chosen defence was therefore not contractual but corporate. Rather than acquire Avison directly, it set up a special-purpose vehicle capitalised at KRW 25.8 billion and let that vehicle buy. The desk reads the intent plainly: confine the acquisition debt in the vehicle and separate it from the parent, so that if the contract dies and the vehicle collapses, EZmedicom itself and the shareholders above it are protected. KRW 25.8 billion of capital controls KRW 100 to 250 billion of contract assets with losses capped at KRW 25.8 billion and gains uncapped, and the party absorbing that asymmetry is whoever lent to the vehicle. This is a hedge for the operator, not for investors.
One leverage benchmark runs across all three. When Affinity acquired 60.1% of Serveone for about KRW 600 billion in 2019, it borrowed KRW 285 billion from the Korea Development Bank in the mid-3% range and converted equity funding into debt, but the collateral was the procurement of the entire LG group. A GPO whose collateral is one hospital’s contract cannot carry that leverage, so the covenants must bind: a loan-to-value ratio of 35% or below, net debt capped at three times earnings, and mandatory disclosure of revenue concentration by hospital. These devices do one thing. They do not raise returns; they distribute the loss, when the read is wrong, across the hospital, the seller and the operator. In the desk’s base case the internal rate of return sits at only 12 to 14%, more than half of it from debt paydown and dividends rather than growth, with effectively no multiple gain. If growth is wrong, returns fall. If the contract is wrong, principal disappears. That is why every clause an architect writes here is tied to the survival of the contract.
What Must Be True
Proposition one: selling 51% while the foundation keeps 49% actually clears the special-relationship test. The most lethal fuse sits inside the Dongha form. With the Dongeun Foundation at 49% through Dongha Industry, and Soonchunhyang Medical Center reported to be running due diligence and the final bid, the question is whether the dominant influence the amended Act describes has been resolved by a share sale at all. Lawyers interpret that reducing a stake below 50% alone is unlikely to remove the violation risk entirely. The falsifier is a special-relationship finding after the 51% is bought: the consequence is not only a trading ban but potential criminal liability, and at that moment the acquired contract becomes unenforceable paper.
Proposition two: the first wave of regulation is the last. The Ministry of Health and Welfare is conducting a fact-finding survey of hospital GPOs through December this year and plans to review further institutional reforms early next year on that basis. The first wave targeted ownership and forced the sales. If the second wave targets margins or transaction transparency, every acquirer that restructured ownership is hit at once; EZmedicom, Hyundai Green Food and Serveone are in the same boat. The falsifier is specific: once a survey publishes a given hospital’s GPO margin, that hospital comes under political pressure to rebid, and a GPO whose customer concentration is effectively 100% enters a revenue-extinction path immediately. Whoever paid the most in wave one becomes the largest casualty of wave two.
Proposition three: the numbers that carry the multiples are the numbers that reach the owner. Avison’s revenue is reported at KRW 364 billion in some sources and KRW 22.9 billion in others, the difference between gross billings and net fees, and that difference distorts every multiple comparison wholesale. Dongha Medical’s revenue has not even been disclosed; its price formed on KRW 7 billion of operating profit alone. And whether the KRW 13 billion of donations continues after the acquisition changes Avison’s real cash flow by a factor of more than two. The first question in diligence is not valuation. It is where the profit was flowing, and whether that channel closes under the new owner.
Proposition four: the Daewoong connection is irrelevant to hospital trust. The industry also reads the Avison acquisition as an extension of Daewoong Pharmaceutical’s hospital distribution network, and argues that concentrating large-hospital supply chains on one platform must change pharmaceutical supply terms and pricing power. The company denies any link to the owner family. But a company whose largest shareholder sits on a pharmaceutical owner’s side, holding a hospital’s purchasing window, stands where the Pharmaceutical Affairs Act, the Fair Trade Act and rebate regulation intersect. The falsifier is a single allegation of product-pushing after integration; the hospital’s trust is this business’s only asset, and one allegation is enough to damage it.
The chain reaction, should these fuses fire in sequence, runs as follows. A second-wave regulatory announcement triggers contract renegotiation; falling profit triggers covenant breaches; suspended dividends and failed refinancing carry the damage past the special-purpose vehicle into EZmedicom’s own liquidity. At that point the platform narrative is demoted to a low-margin distributor, the pool of acquirers narrows, and the multiple steps down. Valuations here do not collapse because the numbers were wrong. They collapse when the 49% that held the contract up and the regulation change direction at the same time, and the numbers merely record the result.
The Architect’s Takeaway
The logic above governs a market where hundreds of billions of won change hands, but it is indifferent to the size of the capital. If you were building your own deal room or platform, the first position to borrow is not EZmedicom’s acquisition strategy but the seat of the Dongeun and Asan foundations. Both sold shares while keeping the volume and the right to choose their vendor. The winners here were not the parties that paid dearly for 51%; they were the parties that held demand and stayed at 49%. Without capital, the first question is not how to buy 51% but in which arena you can become the 49% that holds demand: subscribers for a community operator, user traffic for a developer. Whoever holds it decides who runs the tollbooth, and can take it back.
Second, read what kind of 49% the counterparty holds before discussing price. Before any partnership, investment or employment, ask three questions. Does the party that holds demand remain after the sale? If so, is it close to you, or a stranger who can open another gate at will? Or does it take the money and leave entirely? Price and conditions differ completely across those answers. Applying one price tag to all three is the mistake repeated at MediGood, Dongha and Avison, and the reason asks drifted from KRW 50 billion toward KRW 100 billion.
Third, price what cannot be transferred with deferral. Dongha’s KRW 7 billion was produced by an owner’s relationships, not a system, which is why 10x was attached to a non-transferable asset. Separate income that comes from transferable assets from income that comes from a specific person’s relationship with a specific customer, and if it is the latter, tie more than half of the compensation to that relationship’s persistence, as the Dongha price was split into fixed and contract-linked deferred consideration. Vest a co-founder’s equity against key-customer retention; pay a partner at the milestone rather than up front; if you are the one being recruited, demand the reverse, deferred pay for the relationships you bring, continuing as long as they last. Deferral is not distrust. It is the only honest way to price a non-transferable asset.
Fourth, design differently with and without a hostage. Where the demand holder stays, as at Dongha, attach obligations to the 49%; the logic of the put, that whoever cuts the contract must buy back the shares, costs nothing to write. If you build on someone’s platform, depend on one large customer or co-invest on someone else’s deal flow, you are the GPO: a rule change or a diverted order erases your investment. So take a minimum volume guarantee instead of a promise of exclusivity, a sunk-cost compensation clause instead of an exclusive-supplier lock, and in a co-investment vehicle, link the carry of whoever sourced the deal to the deal’s duration. Where the demand holder takes the money and leaves, as at Avison, there is no hostage, and the downside lives only in the contract’s remaining term, termination compensation and the closing of the leakage channel. Do not cut the price further; double the thickness of the contract. If the counterparty refuses, that is the answer: a party unwilling to share tail risk intends to use you as a buffer, not a partner.
Fifth, an eye for leakage. KRW 13 billion of Avison’s KRW 16.8 billion operating profit left as donations and KRW 0.4 billion of net income remained. Look at where profit exits before you look at the bottom line. Every syndicate or community has a leakage channel: the operator’s management fee, the platform fee, the introduction fee that flows to one person, the service contract under another name. Find it and expose it, or in the arena you design, build it explicitly from the start, with management fee separated from carry and every claim fixed in writing. A structure where leakage is invisible is not a structure without leakage. It is a structure that someone is watching alone.
The last lesson is the coldest. The largest contracts are allocated by families, not auctions; the incumbent lost MediGood to kinship. If you have neither capital nor bloodline, do not try to win on price at a table the family allocates. Find the arenas without a family, and even there compete on structure: while others push the ask up, walk in with deferral, puts and leakage controls. Read where regulation or platform rules will aim next, and vacate the targeted revenue model first. A structure that earns tolls eventually becomes a regulatory target; a structure that earns service fees tends to survive when regulation arrives. The only fair weapon available to someone without capital is time, and time compounds only for those who read structural change first.
Syndicate’s Note
What is sold in this market is not a company but a contract, and the contract’s life is decided not by the price of the 51% but by the character of the remaining 49% and the regulation still to come. The desk files the mechanism in the pattern library’s declassification queue under the working name The Gatekeeper’s Remainder: a regulator forces the owner of a captive supplier to sell majority control, the owner keeps a minority stake and the demand, and the buyer pays an asset multiple for a contract the remainder can revoke. It is the mirror image of Mutual Hostage Architecture. There, buyer and seller are bound to each other after closing and the cash flow behaves like infrastructure; here the binding runs one way, the seller holds the key and the buyer holds the booth, and the Dongha design above is an attempt to manufacture the second direction after the fact.
House Verdict: TRACK. The next twelve months will test the read directly: the identity and terms of Dongha’s final buyer, the ministry’s survey findings and second-wave measures, and whether Avison’s donation channel closes under its new owner. Falsifier: if Dongha Medical’s 51% closes at the full ask with no obligation attached to the foundation’s 49%, and the Soonchunhyang contracts survive both the 2027 deadline and the ministry’s second-round measures intact, then the desk’s read, that the 49% prices the 51%, is wrong. The verdict classifies structural learning value; it is not an investment judgment.
FAQ
What is a hospital GPO (gannapsa) in Korea?
A purchasing intermediary that sits between a hospital and its medical-device and clinical-material vendors, consolidating procurement for a fee or margin. Many were established and owned by the hospital’s own foundation, which is why Korea’s amended Medical Device Act, effective December 2027, now restricts hospitals from procuring through intermediaries that are their special related parties.
Why are Korean hospital foundations selling 51% of their supply subsidiaries rather than 100%?
The regulatory line for a special relationship is drawn at ownership above 50%, so a 51% sale removes the majority holding while the foundation keeps 49% and, in practice, the right to select the hospital’s vendor. The buyer acquires control of the company; the foundation retains the demand. Whether that retained position also constitutes dominant influence under the Act is the open legal question in these deals.
Why did Hyundai Green Food win MediGood Partners over an incumbent supplier like EZmedicom?
Not on price alone, in the desk’s reading. The Asan Foundation sits within the HD Hyundai group and Hyundai Green Food within the Hyundai Department Store Group, a third-degree family relationship that falls outside the Act’s second-degree line while remaining inside the extended family. The foundation describes the process as fair and transparent; the industry calls it a closed league. Both descriptions fit the same structure.
This article is for informational purposes only and does not constitute investment advice, an offer, or a solicitation. Figures are based on public reports and may be estimates.
For more structural insights and deep-dive video breakdowns, visit Structure Syndicate on YouTube.