Four weeks. That is all it took for the world’s best-performing stock market to become a bear market.
In June 2026, the KOSPI crossed 9,300 for the first time in its history. South Korea had leapfrogged both India and the UK to become the sixth largest stock market on the planet. Then, within a month, it was down 30%. $1.3 trillion of market value, gone.
If you only read the headlines, this looks like an AI story that went wrong. It is not. It is a leverage story that happened to be wearing an AI costume. And that distinction matters more to your portfolio than you might think.
The Set Up
Two companies, Samsung Electronics and SK Hynix, made up more than half the KOSPI. On some days, they accounted for nearly 80% of all trading volume in the entire Korean market. That is not a stock market anymore; that is two stocks with a supporting cast.
The reason was simple enough to believe in completely. SK Hynix controlled 58% of the global HBM market, the memory chips that make AI models possible. Samsung and SK Hynix together were projected to generate $1.7 trillion in operating profits between 2026 and 2028. The tax revenue from just these two companies was expected to cover nearly half of South Korea’s national debt.
Every data point pointed the same way. Higher exports. Higher profits. Higher stock prices. Foreign investors, who had avoided Korea for years because of weak corporate governance and the so called “Korea Discount”, started pouring back in. Nearly $17 billion of foreign inflows arrived in the first half of 2024 alone, the strongest year since 2008.
Then came the part that turned a good story into a dangerous one, leverage.
Hong Kong launched the world’s first 2x leveraged ETF on SK Hynix. South Korea followed with leveraged and inverse ETFs on Samsung and SK Hynix directly. Individual investors owned 92% of these products, and most of that exposure was funded with borrowed money, nearly ₩60 trillion by May 2026. Regulators warned about the risk. Nobody listened, because the market kept proving them wrong.
Every Dip Became a Buying Opportunity, Until It Didn’t
Between March and June, the KOSPI fell sharply four separate times, on oil price shocks, US data surprises, policy noise. Each time, it recovered and made a new high. This taught an entire generation of Korean retail investors one lesson, buy the dip, every time, no exceptions.
That lesson works beautifully, right up until the one time it doesn’t.
The break came from an unexpected place, not Korea. Micron reported strong earnings in the US, with customer deals worth over $22 billion. On paper, that should have been good news for the AI trade. Instead, investors asked a different question, is this already priced in. The Philadelphia Semiconductor Index fell nearly 20% from its June peak.
That sentiment crossed the Pacific in a single trading day. On July 13, SK Hynix fell 15.4%, and Samsung fell 10.7%, among their worst single day declines in decades.
This is where leverage stopped being a multiplier of gains and became a multiplier of losses. Leveraged ETFs had to sell aggressively as prices fell, to maintain their target exposure. That selling pushed prices down further. Retail investors who had borrowed to buy got margin called, 1.2 million of them, roughly 3.4% of South Korea’s entire adult population. Their forced selling triggered more margin calls. The very mechanism that had powered the rally on the way up now ran the same loop in reverse, on the way down.
Foreign investors had already pulled nearly $110 billion out of Korean equities through 2026. When the unwind started, there was no buyer left standing.
The final blow came from Korea’s own policymakers. The president publicly flagged instability in financial markets, and the KOSPI dropped 6% that day, erasing $250 billion. A day later, the Bank of Korea raised rates by 25 basis points to fight inflation that had climbed above 3%, right into the middle of a leveraged unwind. Over $600 billion came out of Asian equities as the selling spread across the region.
What Actually Broke
Here is the part investors need to sit with. The AI story in Korea did not fail. Samsung and SK Hynix still control most of the world’s advanced memory chip supply. Demand for AI infrastructure has not reversed. Korean exports hit a record $102.25 billion in June 2026, up 70.9% year on year. Corporate governance is genuinely better than it was two years ago.
What broke was the structure sitting on top of the story. Two stocks carrying more than half an index. Retail investors holding 92% of leveraged products on those same two stocks. ₩60 trillion of borrowed money riding on one trade. That is not a market built to absorb a shock, that is a market built to amplify one, in either direction.
Fundamentals explain why the rally happened. Market structure explains why it unraveled in four weeks.
What This Means for You
We are not writing this because we think you own Korean leveraged ETFs. We are writing this because the mechanics on display in Korea are not unique to Korea, and every investor with equity exposure should be able to recognise them before they show up closer to home.
Three things worth carrying forward.
First, concentration is a risk factor, not a compliment. When a narrative becomes so strong that ownership concentrates into two or three names, or one sector, the story itself becomes the risk. Ask what percentage of your portfolio, or the index you are invested in, sits in the top two or three holdings. If that number is uncomfortably high, you already own the Korea trade, just under a different flag.
Second, leverage does not create returns; it borrows them from the future and charges interest along the way. Leveraged products behave asymmetrically. They amplify gains on the way up and amplify losses on the way down, often through forced mechanical selling that has nothing to do with the underlying business. If you would not be comfortable explaining a leveraged position to us during a 20% single day drawdown, you should not hold it.
Third, a pattern of dips getting bought is not evidence the pattern will continue, it is evidence that positioning has become one sided. The Korean market rewarded dip buyers four times in a row before it punished them once, decisively. The fifth dip always looks exactly like the first four, until it isn’t.
None of this means avoid AI exposure, or avoid Korea, or avoid concentrated markets altogether. It means size these exposures with the understanding that the same forces that drive the upside can, and eventually will, drive the downside with equal force. Good businesses get repriced sharply not because they stopped being good businesses, but because the structure around them was never designed to absorb bad news gracefully.
We would rather you own the story with your eyes open on the structure, than get the story right and get the structure wrong.
