The mint takes five business days. The conversion ratio is fixed — one ADR for 0.1 shares. The depositary is Citibank. The Korean side runs through the Korea Securities Depository. The whole operation settles on something called "administrative procedures," which is finance-speak for fax machines, signed forms, and foreign-exchange declarations processed by humans who leave before six p.m.
SK Hynix just switched on the two-way conversion mechanism between its U.S.-listed American Depositary Receipts (SKHY) and its Korean common stock (000660). The timing matters: the company closed roughly $26.5 billion in ADR issuance in early July. Now the doors are open. The premium on the U.S. side is real. The machinery underneath is exactly what I have spent fifteen years auditing against.
Put it plainly: this is a wrapped asset. Same concept as a bridged token. Same custody logic as a mint-and-burn contract. But it runs on a settlement layer from 1961 and charges its users in days instead of gas.
Code is truth. Intent is fiction. So I audited this the way I audit anything else — by reading the mechanics, not the press release.
For the unfamiliar: an ADR is a U.S.-tradable receipt representing a claim on a foreign company's shares. Citibank holds the actual Korean shares through KSD and issues receipts against them. Investors trade receipts in New York; the underlying equity stays in Seoul.
The conversion mechanism connects the two. ADR holders can redeem into Korean shares. Korean shareholders can deposit stock and receive ADRs. The process runs through Citibank, requires a foreign-exchange filing, passes through brokerage desks on both sides, and takes multiple business days to complete.
Why does this matter? Because SK Hynix is the world's second-largest memory-chip maker — a systemic name in the AI supply chain. Institutional money wants the ADR. The mechanism is the circuit breaker that keeps the two prices from drifting apart. In theory, any meaningful premium gets arbitraged away: buy the cheap side, convert, sell the expensive side.
The theory is elegant. The execution is where the debt lives.
That phrase — "multiple business days" — is not a footnote. It is the risk profile. I scored this mechanism across seven dimensions and landed at 5.75 out of 10: a working product, graded by a man who considers most crypto projects permanent negatives.
What the mechanism actually is
The analysis I pulled categorized the architecture as "concentrated-distributed hybrid." A cleaner formulation: multiple centralized systems joined by slow standards. Each institution — Citibank, KSD, the exchanges, the brokers — runs its own closed ledger. Coordination happens through interbank messaging, regulatory filings, and communications that do not leave a public trail.
A conversion is not a trade. It is a redemption request that moves a claim across registries. The Korean shares do not move in real time. The ADR does not get burned the day you submit. The paperwork flows through an FX declaration to Korean authorities, gets processed by the depositary, and eventually — days later — the position appears on the other side.
I have audited contracts with reentrancy bugs less frightening than this chain of custody. At least in Solidity, every state transition lands on a public chain. In this system, the source of truth is a sequence of signed forms. The comfort is regulatory. The truth is opaque.
One data point tells you everything: the mechanism is "quasi-real-time" at best, with a settlement window measured in days. In 2026, that is not an engineering constraint. It is a policy choice. Intermediaries earn float and spread. Every extra day is a day their balance sheets win. This is not inefficiency. This is the business model, wearing an admin uniform.
The arbitrage lottery
SKHY trades above its Korean equivalent. That premium is what makes the conversion economically interesting. The loop: buy Korean shares, file for conversion, receive ADRs, sell into the premium.
Unit economics depend brutally on that spread. Conversion fees. FX spread. The time cost of locking capital for several days. If the premium is 2% and all-in friction is 0.8%, the arbitrageur nets 1.2% — healthy only until KOSPI moves, or the won-dollar flow turns against the position during the conversion window.
The source analysis flagged this correctly: the mechanism's profitability is sustained only by its own inefficiency. Every successful conversion tightens the premium. When the premium collapses toward zero, arbitrageurs leave. The infrastructure remains — clean, audited, empty.
In 2021, I tracked ownership changes across 1,000 wallets in a famous NFT collection and found 60% wash-trading. The "community" was an artifact of inflation. The same pattern appears here in value form: conversion volume is an artifact of arbitrage opportunity. When the opportunity normalizes, so does the traffic.
Minted nothing, promised everything. That line was written about NFT projects. It fits the premium as well. The premium promises continuous liquidity. The mechanics deliver only as long as the market stays inefficient enough to pay for it.
Where it actually breaks
Credit risk is low — the assets are one-to-one backed. Liquidity risk moderate — the conversion window locks your position. Market risk high — you carry two jurisdictions and two currencies for the days the paperwork runs. Operational risk: highest. Operational risk is the landmine. The FX declaration is my favorite failure node — one regulatory form that must be accurate across two legal systems.
In my 2020 DeFi Summer work, I analyzed 500+ failed transactions when gas spiked and found the network was not the problem. The panic was. Gas fees don't lie. People do. The equivalent failure here: a compliance officer's typo. A slow fax. A weekend. The timeline stretches from "several days" to "next week," and the arbitrageur eats the drift.
The target user is not retail. It is a pension fund with a legal team, or a quant fund with a settlement desk. A retail investor walking into this process gets ground by the paperwork — which is why the only retail presence you will find is on the buying side of the ADR, never on the conversion side.
Regulatory architecture deserves its own note. The mechanism lives under both SEC authority and Korea's Financial Services Commission. Features and burdens. Compliance costs money, and the depositary passes those costs on. Identity, positions, transaction records flow through Citibank, KSD, and broker systems under Korean PIPA and U.S. privacy regimes. Traceable under law, not transparent by design. Soulbound tokens have been a concept for three years because nobody wants a permanent, visible record of their financial history. In this system, the FX declaration is a soulbound token — in government hands, assembled by hand.
What a different rail would look like
Tokenized shares would settle this in seconds. Programmatic escrow. Atomic delivery. No FX filing because a stablecoin pair handles it. I have sympathy for that argument — I called a 90% depeg within 48 hours before Terra collapsed by reading code as machinery, not ideology. Faster rails matter when the code is honest.
But here is the uncomfortable counter-data: the infrastructure that survives is the one regulators already trust. The source analysis scored this mechanism 8/10 on regulatory compliance and 8/10 on macro-policy tailwind. An offshore tokenized share scores zero on both. It does not matter that the code is faster if the code has no legal home where the money lives.
So the real disruption risk is not blockchain. It is RegTech. The highest-value improvement here is automated FX filing, automated AML screening, and workflow robotics that compress conversion from five days to T+1. A software problem with a board-approved budget. The startup that solves it wins the market without touching a chain.
After that, a licensed stablecoin enters the settlement path, and the premium window closes for good. My post-Dencun thesis is that blob space saturates within two years and rollup fees double. Crypto is actively solving that problem. This ADR mechanism has no incentive to solve its own inefficiency. Nobody pays the depositary to be faster. The tax collector does not shorten the line.
Who follows
SK Hynix claims first-mover among Korean large-caps. Samsung and LG watch. The source analysis scores competitive dynamics 6/10, assuming the moat is meaningful. It is not. The moat is regulatory endurance — the willingness to burn legal fees and wait for approvals. Samsung has deeper pockets and could replicate this in months. Differentiation evaporates the day the second Korean giant publishes a conversion schedule.
The quiet variable is whether depositary banks standardize the service. If Citibank packages this as a product for every Korean blue-chip, the platform effect kicks in: per-unit costs fall, liquidity deepens, premiums converge across listings. Bullish for the mechanism. Direct threat to SK Hynix's distinctiveness.
Based on my audit experience, banks monetize infrastructure, not novelty. The question is not whether the mechanism works — it clearly does. The question is whether it stays an exclusive feature or becomes a commodity.
What the bulls got right
Having spent this whole piece dissecting, I will do the honest thing. The bulls are right.
This thing runs. It settled billions in issuance. It is legal in two major jurisdictions. It carries the world's most critical memory-chip flow. And it does so through a process that has never required a community vote, a governance forum, or a token burn.
I have audited projects with polished docs and dead testnets. I have watched promoters present "audits" from firms that would rubber-stamp a haiku. Against that baseline, a regulated, audited, functional mechanism scoring 5.75/10 is a towering achievement. It is boring. Boring is the most valuable property in modern finance.
My own sub-scores look lukewarm: 4/10 on technical architecture, 5/10 on business model, 2/10 on user stickiness. Weighted, 5.75 — "average." Most serious investors would call that a gift. Here is infrastructure that earns its grade by execution, not narrative.
Macro tailwind matters too. The same gravity that pushed MiCA into force in Europe pulls Seoul toward internationalization. Global funds want deeper access to AI supply-chain exposure. A slow, compliant bridge beats a fast, unlicensed one for capital that answers to fiduciaries. I may not love the machinery. But I respect the output: it delivers, on schedule, under law.
More than most protocols I have dismantled can say.
The five-day heartbeat
The ADR conversion mechanism is a five-day settlement tax on global capital flows. It persists because the people who collect the tax control the ledger. The ledger keeps score. Right now, the scoreholder is Citibank, not the investor.
Compression is coming. Not from a chain — from software. Automated FX filings. Straight-through processing. A licensed stablecoin at the settlement layer. The first mover who shortens the window harvests the premium before it dies. The depositary that standardizes the process owns the market. The corporation that ships T+1 conversion becomes the one global funds actually use.
Watch four signals: the premium spread narrowing toward zero; complaints rising around FX processing times; Samsung announcing its own ADR conversion plan; and the first RegTech product that collapses the timeline.
When that last one ships, the five-day heartbeat stops. Bridges that charge in time are relics waiting for retirement. SK Hynix, for all its excellence, just minted a beautiful antique.