If you’d suggested to me a few years ago that a meaningful share of the articles published here during any given week would be dedicated specifically to the daily goings-on in high-bandwidth memory stocks, I would’ve been skeptical.
As a matter of course, I avoid sector-specific editorial saturation. Or at least to the extent that’s possible in a world where to talk about “stocks” is really just to speak about the handful of companies which comprise the lion’s share of market cap.
But as noted here on Wednesday afternoon while editorializing around Micron’s beat and raise, the HBM demand story’s in many respects the load-bearing pillar upon which the AI bull case now rests. And the AI bull case is synonymous with the “broad” market bull case although, as we saw Thursday with Apple’s price hikes and the attendant drop in the stock, this can cut both ways.
With all of that in mind, I wanted to take a moment to highlight some color and a few charts from SocGen’s Asian equities strategy team, which in a new note cautioned that the “exponential rise” in AUM of leveraged single-stock ETFs referencing Samsung and SK Hynix — who between them control around 80% of the global HBM market — is creating a potentially unstable setup in South Korea.
The figures above give you a sense of the ballooning AUM in these ETFs and the extent which AUM in the South Korean products already exceeds that of their Hong Kong-listed counterparts just a month on from launch.
As SocGen’s analysts noted, the issue isn’t the size of the products relative to the underlying. $31 billion in ETF AUM’s meaningless compared to the combined market cap of Samsung and SK Hynix.
Rather, the problem is that those products are on some days responsible for a huge share of underlying turnover in the stocks, where “huge” means as much as 60%.
The figure on the left, above, gives you some context. The figure on the right gives you a sense of the EOD rebalancing impact specifically — it’s been as high as 25% (the lines are seven-day averages).
In their note, SocGen analysts including Frank Benzimra and Rajat Agarwal spell this out in detail, and as far as esoteric explainers go, they do a masterful job of elucidating.
Below, find a short excerpt which I think’s accessible even to the layperson. To wit, from Benzimra and Agarwal:
[These ETFs’] structure embeds a short gamma profile, meaning they must trade with the market direction. In Korea’s volatile, trend-driven semiconductor rally, this has significantly increased the size and frequency of rebalancing flows. As Samsung and SK Hynix move higher, these ETFs are forced to add exposure into strength, while in drawdowns they must de-risk into weakness, amplifying both upside and downside moves. The effect is magnified by rapid AUM growth and retail-driven flows, turning rebalancing into a structural source of demand/supply. As a result, these products have become a non-linear liquidity factor, with rebalancing activity accounting for a meaningful share of daily turnover, reinforcing momentum and increasing short-term volatility.
That’s what I’ve been on (and on and on) about over the past several weeks, a period during which, as noted here at the outset, I’ve written more about the HBM oligopoly than I ever imagined I’d pen in a lifetime.
So, when you bear witness to cringey days like June 23’s session in South Korea — when the Kospi plunged 10% — that’s what you’re seeing. In part anyway. The growing impact of these leveraged retail products referencing two stocks which together comprise almost half of local market cap.




That’s for further elucidating your coverage of these “investment” products, especially in the context of the South Korean market.
This morning I noted that regulators in Seoul had put the brakes on allowing single-stock futures.
Yesterday I as chatting with a colleague about an earlier version of these products. At some point 10 or 15 years ago, there were articles about how “investors” in 2x and 3x levered oil ETFs actually lost money even when they correctly called the market direction. All thanks to the daily reset mechanics you’ve explained to us.
It sounds like you are describing “speed wobbles”:
“The phenomenon occurs when a minor initial disturbance triggers a chain reaction that matches the vehicle frame’s natural resonant frequency, causing the front wheel to repeatedly overcorrect itself out of control.”
Suffice it to say, speed wobbles usually don’t end well.
Can confirm. There’s a speed wobble specific to jeep Wranglers called jeep death wobble. It’s terrifying
It definitely seems that starting around the beginning of 2026, I’m noticing bigger reactions to smaller “initial sparks”.