The best-performing index in the world

Five days ago, the Kospi was up 47 percent for the year.^[6] Korea's stock market capitalization had surpassed France's. Samsung's share price had nearly quadrupled since the start of 2025. SK Hynix had gone up six-fold. Global funds poured more than $7 billion into Korean equities in two months. Retail investors were sharing memes — literally, memes — about semiconductor stocks with captions like "No time to explain, get in."^[7]

The top-performing ETFs in the country were all leveraged semiconductor products: TIGER Semiconductor TOP10 Leverage, KODEX Semiconductor Leverage, TIGER 200IT Leverage.^[11] Retail investors were net buyers of 821 billion won in TIGER Semiconductor TOP10 in February alone.

The Kospi wasn't an equity index. It was a leveraged bet on two memory companies that the world's most aggressive retail investor base had convinced itself could only go up.

Then the Strait of Hormuz closed.

What 12 percent looks like

On Wednesday, the Kospi fell 12.1 percent to 5,093.54 — the worst single-session decline in the benchmark's history, eclipsing even the crash that followed September 11, 2001.^[1] Samsung Electronics fell nearly 12 percent. SK Hynix dropped about 10 percent. The two-day drawdown was the steepest since 2008.^[4]

The won, meanwhile, briefly pierced 1,500 to the dollar in offshore trading — a level last seen in March 2009, during the global financial crisis.^[10] It stabilized back below 1,500 after what we can only assume was some vigorous intervention by the Bank of Korea, though the central bank's official statement used the phrase "dollar liquidity remains ample," which is the kind of thing central banks say when dollar liquidity is in question.

The immediate catalyst is obvious: the Iran war. But the question worth asking is why Korea, specifically, took the worst hit of any major market on earth. The Nikkei was down 5 percent. The Hang Seng fell 4 percent. The Kospi fell 12. Why?

The concentration problem

South Korea imports 98 percent of its fossil fuels. Roughly 70 percent of its crude comes from the Middle East, most of it transiting the Strait of Hormuz.^[2] That's the energy dependency story, and it's real but not unique — Japan imports 95 percent of its crude from the Middle East and managed a less spectacular decline.

What made Korea different was the index itself.

The Kospi's 2026 rally was overwhelmingly a semiconductor story. Samsung and SK Hynix together account for a dominant share of the benchmark's weighting. When we say "the Kospi was up 47 percent," what we really mean is "Samsung and SK Hynix were up a lot, and the Kospi came along for the ride." The index had become a single-factor bet on memory chips, dressed up as a diversified equity market.

SK Hynix posted record results in 2025: revenue of 97.1 trillion won, operating profit of 47.2 trillion won, a 49 percent operating margin — beating Samsung in annual profit for the first time in history.^[8][9] HBM revenue more than doubled year-on-year. The numbers were, genuinely, extraordinary.

But extraordinary numbers attracted extraordinary positioning. And extraordinary positioning in a concentrated index creates the conditions for extraordinary reversals.

The leverage problem

Here's what the meme accounts didn't mention. Leveraged ETFs amplify returns in both directions. A 2x leveraged semiconductor ETF that was up 35 percent in February is, mechanically, facing a drawdown roughly twice the magnitude of the underlying when the trade reverses. Retail investors who bought TIGER Semiconductor TOP10 Leverage at the peak are now looking at losses that make the 12 percent index decline seem gentle.

The margin loan data tells its own story. Korean retail investors — the famous "ant investors" who dominate daily turnover on the Kospi — had been adding leverage throughout the rally. Individual investors made net purchases of 119 billion won on Tuesday alone, buying the dip as foreign investors sold. The foreigners, it turns out, were the ones who had the right instinct.

Foreign outflows accelerated through the week as the geopolitical situation deteriorated. The dynamic is self-reinforcing: foreign selling weakens the won, which makes Korean assets less attractive to foreign holders on a hedged basis, which triggers more selling, which weakens the won further. The FX-equity doom loop is not a theoretical construct. It is what happened on Wednesday.

The structural vulnerability

The Korea story is ultimately about what happens when an index becomes a trade and then the trade's assumptions get invalidated.

The assumptions were:

1. AI demand for HBM memory would keep growing. (Still true.)
2. Samsung and SK Hynix would capture the margins. (Still plausible.)
3. Korea's energy import dependency was a background risk, not a foreground risk. (No longer true.)
4. The won would be stable enough to make unhedged Korean equity exposure tolerable for foreign capital. (Not this week.)

Assumptions one and two didn't change. Assumptions three and four did. And because the index was so concentrated in a single theme, there was no diversification buffer. The Kospi had no place to hide because the Kospi was the semiconductor trade, and the semiconductor trade doesn't hedge oil shocks.

What to watch

The Kospi surged past 6,300 six days ago.^[5] It closed Wednesday at 5,093. That's a 19 percent peak-to-trough drawdown in less than a week.

The fundamental case for Korean memory companies hasn't changed — HBM demand is real, pricing power is intact, and SK Hynix's margins are the envy of the semiconductor industry. What changed is that a concentrated equity market in an energy-importing country with an unstable currency got hit by an energy shock, and all the leverage that juiced the rally amplified the selloff.

The won at 1,500 is a number that forces decisions. The Bank of Korea has to choose between defending the currency (tightening into a potential recession) and letting it slide (which accelerates capital outflows and makes the oil import bill even more painful). Neither option is good. Both are available.

Korea's "AI index" was the best story in global equities for two months. It may well be the best story again — the underlying business fundamentals haven't disappeared. But this week was a reminder that an equity market is not a sector bet, a country is not a company, and the Strait of Hormuz doesn't care about your HBM revenue forecast.^[12]

The ants are still buying. We suppose someone has to.