Another day, another absolutely harrowing session in the casino that is South Korea’s domestic equity market.

The local benchmark, you’re reminded, is hostage to the fate of Samsung and SK Hynix, which between them comprise better than half of overall market cap. Samsung and SK Hynix, in turn, were hijacked in recent months by a suite of leveraged retail products that routinely account for better than half of underlying turnover.

In simple terms, the structure of those products means they buy into strength and sell into weakness, amplifying directional moves. The result: Outsized swings and elevated volatility. By the time South Korean regulators set about addressing the situation (which they in part caused by approving single-stock leveraged ETFs in the first place), it was too late.

It’s no longer unusual for local equities to move 5%, 8% or even 10% in a single day. As one Singapore-based strategist put it last month, when the nation’s benchmark suffered one such dramatic decline, “[Time was] I would not be here in this room if the Kospi drops this much in a day, I would be on the phone talking to clients [or] tearing away at my laptop trying to write something. Now it’s a normal day.”

In that sense, Tuesday was just another “normal day.” The Kospi fell nearly 11%. As the simple figure below shows, Tuesday marked the third time in 2026 that the benchmark dropped 10% in a single session.

If you don’t understand how insane that it, I won’t bother trying to (re)explain it other than to remind you that South Korea isn’t some backwater — it’s a quasi-DM.

The losses were mirrored in (and caused by) double-digit swoons for Samsung and SK Hynix. The former shed 13.4%, the latter almost 15%.

A month ago, SK Hynix was worth almost $1.4 trillion. Now, after plunging nearly 50%, it’s a $785 billion company, trading on a sub-4 forward multiple.

I imagine they’ll be some US retail investors looking to buy the newly-issued ADRs on weakness. I won’t weigh in on the merits of buying this particular dip. But I will gently note that multiples must be considered in context, which is to say if you don’t know why multiples tend to be low in the memory space, you probably shouldn’t be dabbling in it.

Relatedly, earnings estimates for these stocks are inflated by assumptions about the sustainability of AI capex and the notion that the current supply-demand imbalance will persist more or less indefinitely. Those assumptions could be wrong for any number of reasons. Demand for AI could disappoint, for example, or supply could catch up more quickly than anticipated.

On that latter point, news that China’s making progress on the ultraviolet lithography front probably didn’t help the mood on Tuesday in South Korea. And CXMT’s raucous public debut on Monday raised the specter of intense competition from China’s homegrown DRAM champion.

SK Hynix reports earnings on Wednesday.