Korea isn't diversification: it's one AI bet in a different currency

Would you call a portfolio diversified if half of it moved on the same earnings call? That is roughly what an investor gets when buying South Korea as a way to "spread out" emerging-market risk. The country is no longer a broad economy with a stock market attached. It is, in our view, a listed bet on memory chips.

The numbers say the index is two companies

The recent history is a case study. In early June, TradingKey reported that Samsung Electronics and SK Hynix together account for over 50% of the Kospi's weight, and that nearly three-quarters of the index's gains that year came from the two. That day, the Kospi fell 5.54% after a disappointing AI guidance from Broadcom, a U.S. company. Nothing happened in Korea, yet Korea took the hit.

The pattern has repeated. On July 16, SK Hynix fell 11.53% and the Kospi dropped 6.37%. According to MLQ's account of that session, the two chipmakers plus 16 single-stock leveraged and inverse ETFs made up 96.3% of total Kospi trading volume. The Kospi had triggered its sell-side "sidecar" pause 37 times in 2026, against just six from 2000 through 2025. Then, in mid-October, Bloomberg's headline described Korean stocks tumbling 8%, with the Kospi about 21% below its September high.

That is not the profile of a diversifier. A diversifier should zig when your U.S. tech holdings zag. Korea has been zigging with them, only harder.

The crowd makes the theme louder

Concentration is one problem; who holds it is another. CNBC reported in May that foreign investors dumped $13.2 billion of Korean equities in a single week, while retail investors piled in, often through margin trading and leveraged ETFs. Citi warned of "exuberance" and took profits on half its Korea position.

We think that is the underrated detail. When retail leverage sits on top of two stocks, price moves become mechanical. A Kiwoom Securities analyst, Han Ji-young, said as much about July, attributing the drop to heavy concentration and fund-flow volatility rather than weaker demand signals. Mechanical selling can reverse quickly, as the rebound into a technical bull market in August showed. But it also means the swings are violent in both directions, and timing them is not something an average saver does well.

What a sensible reader should do

None of this means Korea is a bad market or that the AI memory cycle is over. Analysts quoted in the July coverage called that selloff a cyclical correction, and Morningstar's Lorraine Tan said the upcycle remained intact while valuations had run ahead of fundamentals. Those are forecasts, not facts, and they can be wrong in either direction.

The point is about labeling. If you buy a Korea ETF, call it what it is: a thematic position in memory chips, sized like one. Check how much AI you already own through U.S. index funds, where semiconductors are roughly 20% of the S&P 500, per one strategist cited by MLQ. Then ask whether a third helping is what you actually want.

Our verdict: Korea is the purest AI trade on the board, and that is precisely why it fails as diversification. Investors who want it deliberately, with a small position and a stomach for 6% daily drops, can make a case. Those who bought it for balance should look again. This is analysis, not personal advice, and your own goals and tax situation matter more than any ticker.