Why
A digital-asset custodian announced the close of its Pre-A round on 2026-08-24. The strategic investors named are Sh Suhyup Bank, iM Bank and HFR (KOSDAQ 230240), and the company states the position plainly: Korean banks now hold about 30%, and close to 40% counting SK Securities, an existing shareholder.
Four days earlier the news was that Suhyup had taken 14.95% — a number that reads as deliberate, because the Banking Act bars a bank from holding more than 15% of a company that is not its subsidiary, and crossing that line brings approval and consolidation with it. (Worth confirming against the statute; the card should not lean on the reading harder than the evidence allows.)
Put the two together and the shape appears. If banks hold ~30% and one of them holds 14.95%, the other is sitting just under the same ceiling. Each stake individually is below the line that triggers review. The bloc is not.
That is this catalogue's most repeated shape, arriving somewhere new. stake-concentration found it in a validator set — n operators that share one correlated failure. aqua-shared-liquidity found it in an order book — three positions quoting one balance. dkg-resharing found it in a key committee. Here it is a cap table, and the correlated class is "Korean banks", which is about as correlated as a class gets: same regulator, same funding model, same incentive to protect the same incumbent position.
How it works
The gap: approval is per holder, control is aggregate
| Level | What is reviewed | Who reviews it |
|---|---|---|
| One bank at 14.95% | Below the subsidiary threshold — no approval triggered | Nobody, by design |
| A second bank at ~15% | Same, reviewed separately if at all | Nobody, by design |
| Banks at ~30% combined | Nothing | No process looks at the sum |
| ~40% with a securities firm | Nothing | — |
The ceiling does its job perfectly at the level it operates on. It was never designed to see a class. Each holder is measured against the line alone, so a correlated group can assemble a controlling position out of individually-compliant pieces without any single filing ever being wrong.
Why this is not an accusation
Nothing here suggests coordination, and the card should not imply it. Two banks investing in the same infrastructure company at the same moment is exactly what you would expect if the sector reached the same conclusion independently — which is the point. Correlated behaviour does not require a conspiracy; it requires a shared incentive, and a shared regulator plus a shared threat model supplies one. The stake-concentration card makes the same move about validators sharing a cloud region: nobody colluded, the correlation was structural.
What is actually at stake
The company's own framing is that it wants to be the infrastructure connecting institutional finance to digital assets — corporate market participation, stablecoins, spot Bitcoin ETFs. If that succeeds, this is the custodian sitting under a whole national market's institutional flow, and its shareholders are the incumbents that flow would otherwise route around.
| If the custodian becomes the standard rail | Then the ~40% holder class is |
|---|---|
| Corporate digital-asset accounts | The banks those corporates already bank with |
| Stablecoin issuance and redemption | The issuers' most likely competitors |
| Spot ETF custody | The distributors of the competing products |
None of those is improper. All of them are worth being able to see, which is the measurable part.
The measurement
This is the stake-concentration method applied to a cap table, and it needs no privileged data:
- Take Korean digital-asset infrastructure firms — custodians, VASPs, stablecoin-adjacent issuers.
- Reconstruct shareholding from public filings, press releases and company announcements.
- Classify each holder into a correlated class: bank, securities firm, non-financial strategic, VC, founder.
- Compute a concentration index per class, and the largest single class share.
- Compare that against whatever threshold actually triggers review.
Output: the fraction of a piece of national financial infrastructure held by one correlated class, next to the threshold that class never has to cross individually. Nobody publishes that number, and the whole argument is one table wide.
Open questions
- Is there any rule that aggregates? Korean financial regulation has concepts for acting-in-concert and for 동일인 in other contexts. Whether any of them reaches unrelated banks holding the same target is the question that decides whether this is a gap or just something looked at elsewhere. Answer this before the card claims a gap exists.
- What are the actual stakes? The ~30% is the company's own statement, not a filing. iM Bank's exact holding is not public here.
- Does it matter at 30%, or only at board level? Influence is not linear in percentage. Board seats, veto rights and information rights would tell more than the number does, and none of those is in the announcement.
Where it lands in Jayverse
- Token: add a correlated-class concentration check to JYVE distribution. Per-wallet caps mean nothing if related wallets (team, advisors, one fund's sub-accounts) each sit just under the line; reconstruct holdings and classify them the way this page classifies bank holders, and cap the class, not only the address.
- Verex: classify market-maker and LP counterparties into correlated classes for the risk engine. Several correlated bot addresses each under a per-account exposure limit can still be one correlated risk; compute a concentration index per class, not just per account.
- Auditor: adopt "reconstruct holdings, classify into correlated classes, compute the largest class share" as a standing method. Publish it alongside per-holder numbers for any Jayverse token or governance system, the same gap this page finds in bank ownership.
Key expressions
| Expression | 뜻 · 쓰이는 자리 |
|---|---|
| statutory ceiling | 법정 상한(선) · 은행법 등에서 정한 보유 한도를 말할 때 · "took a stake just under the statutory ceiling" |
| acting-in-concert | 공동행위(합의된 공동 보유) · 여러 주체가 사실상 하나처럼 움직이는지 볼 때 · "(acting-in-concert, 동일인)?" |
| cap tables | 지분 구조표들 · 주주 구성을 정리한 표를 가리킬 때 · "reconstruct the cap tables and compute the largest correlated-class share" |
| approval triggered | 심사·승인이 발동되다 · 특정 지분율을 넘으면 규제 검토가 시작될 때 · "Below the subsidiary threshold — no approval triggered" |
| sitting just under | (기준선) 바로 아래에 머물러 있다 · 규제 문턱을 아슬아슬하게 피하는 상황 · "sitting just under the same ceiling" |
| bloc | (이해관계가 같은) 집단·블록 · 개별로는 규제 아래지만 합치면 큰 세력이 될 때 · "The bloc is not" |
| correlated class | 상관관계가 있는 부류(같은 유인을 공유하는 집단) · 은행들처럼 같은 규제·동기를 공유하는 집단 · "same regulator, same funding model, same incentive" |
| incumbents | 기존 기득권자(기존 사업자) · 새 흐름이 위협할 대상을 가리킬 때 · "the incumbents that flow would otherwise route around" |
| route around | ~을 우회하다 · 새로운 흐름이 기존 경로를 피해 갈 때 · "flow would otherwise route around" |
| lean on | (해석을) 과도하게 의존하다·밀어붙이다 · 근거보다 앞서 결론짓지 말라는 경고 · "should not lean on the reading harder than" |
| Banking Act | 은행법(한국 법률) · 은행의 비계열사 지분 보유를 15%로 제한하는 규정, 초과 시 승인·연결 대상 · "the Banking Act bars a bank from holding more than 15%" |
| HFR | 코스닥 상장기업(종목코드 230240) · 이번 프리A 라운드에 참여한 전략적 투자자 중 하나 · "iM Bank and HFR (KOSDAQ 230240)" |
| KOSDAQ | 한국 코스닥 시장 · 중소·벤처기업 전용 증권거래소, 기업의 상장 시장을 가리킬 때 · "HFR (KOSDAQ 230240)" |
| SK Securities | SK증권 · 이 커스터디 기업의 기존 주주인 한국 증권사, ~40% 지분 계산에 포함 · "close to 40% counting SK Securities" |
| VASPs | 가상자산사업자(Virtual Asset Service Provider) · 거래소·커스터디 업체를 가리키는 한국 규제상 법적 분류 · "custodians, VASPs, stablecoin-adjacent issuers" |
| consolidation | 연결(회계상 종속회사로 편입되어 재무제표를 합산하는 것) · 지분율 초과 시 은행법상 발생하는 결과 · "crossing that line brings approval and consolidation with it" |