The Second Door

SK Hynix lists on the Nasdaq today as SKHY, in the largest first-time US share sale a foreign company has ever done. Nearly every headline calls it an IPO. It is not one. The company was already public. What is really happening is a memory maker opening a second door onto the deepest pool of capital on earth, and raising more than twenty-six billion dollars of fresh money to build fabs. How that works, and why it is happening now, says more about the AI buildout than the headline number does.

A record fell today. SK Hynix listed on the Nasdaq under the ticker SKHY, in an offering that priced at roughly twenty-six and a half billion dollars after landing more than seven times oversubscribed. That clears the old marks held by Alibaba and Saudi Aramco and makes it the largest first-time US share sale a foreign company has ever run.

Be precise about which record this is. It is not the biggest US listing outright. SpaceX was far larger last month. It is the biggest any foreign company has ever done, and that distinction is the tell. A foreign issuer coming to US markets is the ADR story, not the IPO story, and those are two different machines.

Almost every headline is calling it an IPO. It is not one. Understanding why it is not, and what it is instead, tells you more about where the AI buildout stands right now than the record does. So this week we took the deal apart and put it back together, because once you see how the machinery works, you see the thing underneath it.

The thing everyone gets wrong

SK Hynix did not go public today. It is already public. It has traded on the Korea Exchange for decades, in won, Korea's currency, and it even has depositary receipts listed in Luxembourg. A US investor could always own it in theory. In practice, buying shares on a Korean exchange, in a foreign currency, through infrastructure most American brokerage accounts do not reach, was enough friction that almost nobody did.

So the first correction. This is not a company arriving in public markets. It is a company that is already one of the most valuable in the world adding a second front door, one that opens onto the deepest pool of capital on the planet. That door is the ADR.

What an ADR actually is

An American Depositary Receipt solves a simple problem. Americans want to own foreign companies, but foreign shares live on foreign exchanges, in foreign currencies, under foreign settlement rules. That is a lot of friction.

So a US bank steps in as a middleman. It holds the foreign shares, and against them it issues receipts that trade here, in dollars, during US market hours, in a normal brokerage account, like any American stock. You hold the receipt. The bank holds the actual share. You get the economics of owning the foreign company without ever touching the foreign market. That receipt is the ADR. For SK Hynix, one Korean share is split into ten of them, which is why the Nasdaq price looks like roughly a tenth of the Seoul price. That ratio is just plumbing.

Here is the fork in the road that separates this deal from a simple currency wrapper.

Two kinds of ADR, and only one raises money

Some ADRs just wrap shares that already exist. The bank buys up ordinary shares already trading abroad and issues receipts against them. No new stock is created, and the company raises nothing. It is pure access, a translation layer. Most ADRs you have heard of work this way.

SK Hynix is doing the other kind. This is a capital-raising offering. The company is issuing about 17.79 million brand-new shares, roughly 2.5 percent of itself, specifically to sell into this deal. That is real new stock. It raises real money. And because new shares now exist, it dilutes existing holders slightly, by that same 2.5 percent.

Do not let the phrase fool you into thinking nothing was raised. This is the fundraising of an IPO grafted onto a company that already trades. That hybrid is exactly what this kind of ADR is built to do, and it is why the "just a wrapper" instinct is wrong.

Where the money goes

Every dollar of the proceeds is already spoken for, and the destination is the argument in one line.

Read that list again. Not one dollar goes to selling shareholders. All of it goes into the ground, into the specific capacity required to make more high-bandwidth memory, the exact product that is sold out through 2027 and that every AI accelerator now depends on.

Sit with the strangeness of that. SK Hynix is sitting on tens of trillions of won in net cash. It does not need the money in the ordinary sense. It is issuing new equity anyway, and pointing all of it at capacity.

The clearest read on how real the AI memory shortage is may be that the market's strongest supplier is willing to dilute its own shareholders to pour concrete faster.

The misconception worth killing

The stock drifted down in the weeks before the listing, and the offering was quietly trimmed with it, from roughly twenty-nine billion to about twenty-six and a half. The lazy take is that this is a discount engineered to hand US buyers a sweetheart entry. That gets the incentives backwards.

Because the proceeds fund the company's own expansion rather than cashing out insiders, a low price hurts the issuer. Every share sold cheap is capital left on the table for the fabs. What SK Hynix and its four banks actually want is a price that cleanly fills a book this enormous without leaving it undersubscribed. That the book still came in more than seven times oversubscribed at the trimmed price tells you which way the demand was leaning. That is not a discount handed to anyone. It is the physics of placing the largest foreign-company listing on record without it wobbling.

Why they are really doing this

Here is the logic. SK Hynix has traded at a lower multiple of its earnings than its US-listed peer Micron, despite arguably holding the stronger position in high-bandwidth memory, the segment that matters most. The stronger player has carried the cheaper valuation.

That gap has a name: the Korea discount, the persistent tendency of Korean companies to trade below comparable Western peers, partly because the investor base was fenced off behind local market infrastructure. When the same company suddenly sits on a US screen, one line above or below Micron, in the same currency, in the same brokerage account, that gap becomes very hard to look at. The whole re-rating case rests on a simple mechanical bet: put the asset where the deepest capital can see it clearly, and the discount has a harder time surviving.

The mechanical catalyst underneath the story

A US listing opens the door to US index inclusion, and inclusion is not a matter of opinion. When a stock enters a major benchmark, every passive fund that tracks that benchmark has to buy it, on a schedule, regardless of what anyone thinks it is worth. That is mechanical demand, and it is a large part of why a company would want to be on a US exchange at all.

The near-term prize analysts keep pointing at is the Philadelphia Semiconductor Index, the benchmark that the big chip funds track. A US-listed memory leader of this size is a natural candidate, and inclusion there would pull in passive money that could never reach the Korean line. Broader index entry, the kind that draws the largest passive pools, tends to come later and on a set calendar rather than on day one, so the mechanical bid builds over months, not in a single session. The point is simpler than the timing: putting the asset on a US exchange is what makes it reachable by pools of capital that were structurally walled off before.

The part that outlives the ticker

As we wrote last week, when a thing is scarce, takes years to build, and has a customer with a signed order and no substitute, the maker of that thing sets the terms. The same logic runs through capital itself. If this were only about one Korean chipmaker, it would be a big deal and a small story. It is bigger than that.

Japan's Kioxia has already said it will do its own US listing next spring. Samsung is fielding a surge of questions about doing the same. What listed today is the first and largest move in a broader migration. The Asian memory makers, the companies that physically supply the memory the entire buildout runs on, are systematically crossing the Pacific to raise capital in US markets, because that is where the money to fund the buildout actually lives.

The risk runs the other way too, and we hold both. A re-rating is a scenario, not a promise. If demand cools or capital pulls back, a stock that raised into strength can give it back, and the same US screen that invites a re-rating up can price one down just as fast. The discipline is to understand the machinery, not to bet on the mood.