01
On 10 July, SK Hynix officially made its debut on Nasdaq.
Its offer price was $149, but the stock opened at $170, climbed as high as $177 intraday, and finished its first session at $168, up 12.76 per cent. That gave it a market capitalisation of roughly $1.22 trillion, and at one point put it nearly $200 billion ahead of Micron.
Yet the share price pop was not the most striking part of the story. The real shock was the size of the deal: $26.5 billion. To put that in perspective, it surpassed Alibaba’s $25 billion US listing in 2014 to become the largest-ever US IPO by a foreign company. On a global basis, it ranks behind only SpaceX and Saudi Aramco, making it the third-largest IPO in history. The issue was more than seven times subscribed, with top-tier funds such as Baillie Gifford and Coatue scrambling for allocations.
But SK Hynix did not come to market through a standard IPO. Instead, it used an ADR structure. ADR stands for American Depositary Receipt. Put simply, a US bank holds the foreign company’s underlying shares, while investors in America trade receipts rather than the shares themselves. In SK Hynix’s case, every 10 ADRs represent one Korean ordinary share, denominated in US dollars and traded during normal US market hours.
That raises an obvious question. SK Hynix was already listed in Korea and doing perfectly well. Its share price had risen 770 per cent over the past year. By the end of June, its market capitalisation had overtaken Samsung’s, making it the most valuable listed company in Korea. Its first-quarter operating margin stood at 72 per cent, comfortably ahead of both Nvidia and TSMC. It controls roughly 57 per cent of the global HBM market and is the single most important supplier behind Nvidia’s AI chips. It also has KRW 35 trillion in cash on the balance sheet.
If it is not short of money or profile, why list again in the United States?
02
The most obvious answer is scale. The US equity market and the Korean market are simply not in the same league.
US equities are worth about $70 trillion in total, versus roughly $4.4 trillion for South Korea’s market, only about 6 per cent of the size. The difference is less a pond versus a lake than a pond versus an ocean. No matter how dominant SK Hynix may be at home, the pool of capital it can reach in Korea is nowhere near what is available in US markets.
More importantly, a large number of American institutions cannot buy Korean shares directly.
Pension funds, mutual funds and sovereign wealth funds manage enormous pools of capital, and many of them are allowed to hold only securities listed in the United States. When SK Hynix traded solely in Seoul, that money simply could not get in.
In other words, it was not that those investors did not want to own SK Hynix. In many cases, they were not allowed to. Now that the company is listed on Nasdaq, that barrier has been removed.
But eligibility is only the first step. The more important consequence is that, once listed via ADR, SK Hynix becomes a candidate for inclusion in a range of indices, opening the door to passive flows.
On the current timetable, the company could eventually be added to indices such as the KOSPI 200, the Philadelphia Semiconductor Index and the Nasdaq-100. That matters because ETFs and pension funds tracking those benchmarks would then have to buy the stock. The Philadelphia Semiconductor Index alone could bring around $3.8 billion in passive demand, while the capital behind the Nasdaq-100 is larger still. There are also potential rebalancings by FTSE, MSCI and other global index providers, which together could translate into additional allocations worth tens of billions of dollars.
These flows are still prospective for now, but if inclusion happens, the money comes in almost automatically. That is how the mechanism works.
Beyond market access, there are two other practical considerations.
The first is funding. SK Hynix has announced a KRW 100 trillion capacity expansion plan. Raising dollars makes strategic sense if it expects to build plants overseas and spend directly in dollars rather than converting currencies along the way. Chairman Chey Tae-won has already said the group remains open to building chip operations on US soil.
The second is proximity to customers. Nvidia, Google, Amazon and Microsoft are all in the United States. Its biggest clients are there, and HBM orders are already booked out to 2028.
So this is not a story about a company that could no longer thrive in Korea. It is a story about turning on a much larger tap.
03
SK Hynix is not the first Asian chip champion to take this route.
TSMC issued ADRs back in 1997 and has traded on the New York Stock Exchange for nearly three decades. In the industry, the broad view is that SK Hynix’s US listing was inspired in large part by the success of TSMC’s ADR programme. TSMC’s ADRs have long enjoyed strong liquidity and have consistently traded at a premium to the Taiwan-listed shares, making them highly attractive to global investors.
That premium tells an important story. Over the past decade, TSMC’s ADRs have traded at an average premium of about 7.4 per cent. At the height of the AI rally, the premium briefly reached 30 per cent. It is currently around 16 per cent. In other words, investors in the United States are still willing to pay about 16 per cent more for essentially the same underlying asset than investors in Taiwan.
Why does that happen? There are two main reasons.
First, supply is constrained. Converting Taiwan-listed shares into ADRs requires regulatory approval, which limits the amount of new ADR issuance. Supply cannot expand freely.
Second, index mechanics matter. TSMC’s ADRs are included in benchmarks such as the Philadelphia Semiconductor Index, so ETFs tracking those indices must buy the ADRs rather than the local shares. Demand remains firm, while supply remains tight, and the premium becomes a structural feature rather than a one-off anomaly.
UBS has recommended that clients go long SK Hynix ADRs and short the Korean-listed shares. The logic is straightforward: ADRs offer lower frictions and broader global access, while converting Korean shares into ADRs still requires approval from Korean regulators, constraining supply and potentially supporting a premium. TSMC’s roughly 16 per cent premium versus its Taiwan shares remains the clearest precedent.
But TSMC’s experience also carries a warning: premiums are not fixed. They fluctuate. When enthusiasm among local Taiwanese investors runs hot, the Taiwan-listed shares can rise faster and the premium narrows. When foreign capital leads the move, the ADRs tend to outperform and the premium widens. And shorting the ADR while going long the local shares can be a dangerous arbitrage trade, because the premium does not necessarily mean-revert.
Some analysts have explicitly warned that global investors have already recognised TSMC’s technological leadership and have assigned its ADRs a structurally higher valuation. If so, the premium may never return to its old range.
TSMC’s three-decade experience shows that the ADR model can work. But the durability of TSMC’s premium is inseparable from the strength of its moat. The real question is whether SK Hynix has the same kind of staying power.
04
Can SK Hynix sustain an ADR premium in the way TSMC has? In the end, the answer depends on one question: how difficult is it to replace what the company does?
That is where the case becomes harder to make.
This is not to say SK Hynix is weak. A 72 per cent operating margin and a 57 per cent share of the global HBM market are formidable numbers. But memory chips are, by nature, a cyclical business. That is very different from TSMC’s foundry model, where customers are often deeply dependent on its manufacturing capabilities.
The semiconductor industry typically moves in cycles lasting six to ten years, and each major upcycle tends to begin with an explosion in downstream demand: personal computers in the 1990s, feature phones and laptops in the 2000s, smartphones in the 2010s, and now AI.
The AI computing boom has pushed memory into a super-cycle. Demand is growing at 38 per cent, more than four times the pace seen in traditional cycles. In 2026, global memory-chip output is expected to reach as much as $550 billion, up 42 per cent year on year, overtaking foundry to become the largest sub-sector in semiconductors.
But cycles always reassert themselves. The previous downturn is still fresh in investors’ minds. Between 2022 and 2023, DRAM prices were cut in half. Samsung Electronics saw quarterly profits collapse by 95 per cent, and SK Hynix recorded its first quarterly loss in a decade. The history of memory chips has always been a loop from windfall profits to deep losses. The question is whether this time is really different.
Perhaps it is. But investors should not ignore the supply side. Samsung, SK Hynix and Micron are expected to invest more than $120 billion combined in capacity expansion over the next several years, with planned new output set to come on stream in a concentrated wave by the third quarter of 2027. Global memory supply is going to rise sharply. With that much money being deployed, additional supply is not a hypothesis; it is a matter of time.
There is another variable too: China.
ChangXin Memory Technologies has already cleared its domestic listing review and is seeking to raise RMB 29.5 billion in what could become the largest A-share IPO of the year. It currently holds about 8 per cent of the global DRAM market, ranking fourth worldwide. Its technology still trails the leaders by around two to three years, and its combined HBM yield is only about 25 per cent, with 99 per cent of shipments still going into consumer-grade products.
In the short term, the threat is limited. But the determination and capital behind China’s import-substitution drive should not be underestimated. Apple is reportedly lobbying the US government for a special licence to buy memory chips from ChangXin. If even Apple has to lobby Washington for permission, that in itself suggests ChangXin’s products can no longer be dismissed.
05
In the short run, SK Hynix’s US listing looks like a success. The deal was more than seven times subscribed, the stock rose 12.76 per cent on day one, and global capital has clearly cast its vote with real money.
But over the longer term, the company still faces the old questions that have always haunted the memory industry. Can it break free from the sector’s cyclical destiny? Is its moat deep enough? And will rising Chinese challengers change the competitive landscape?
TSMC’s experience over nearly thirty years shows that the ADR route can work. But TSMC’s moat and SK Hynix’s moat are not the same thing.
So how much more is a company worth simply because it switches exchanges?
In the short term, an ADR premium and index-driven passive inflows can clearly support a higher valuation.
In the long term, however, what determines a company’s value is not the name of the exchange on which it is listed, but how hard it is to replace.