Regulatory actions in South Korea trigger a stampede; profit-taking hits the AI bull market for memory chips, with sector index down 7% intraday and SanDisk plunging over 10%.
```
One of the most crowded sectors in AI trading is experiencing a sharp correction.
During intraday U.S. trading on Thursday, storage chip stocks collectively plunged. The Philadelphia Semiconductor Index (SOX), which tracks overall semiconductor performance, once dropped more than 4%, and the storage chip index fell as much as 7%.
As for individual stocks, SanDisk (SDNK) fell more than 10% intraday, Western Digital (WDC) dropped nearly 9%, SK Hynix ADR (SKHY) fell nearly 9%, Seagate Technology (STX) dropped more than 8%, and Micron Technology once fell more than 6% intraday. Over the last two sessions, its cumulative decline exceeded 10%, and it has retreated over 30% from the intraday high on June 22.
This round of correction also signals that the storage chip sector, which has been most sought after by the AI boom this year, is beginning to face market repricing. As South Korea’s regulators tighten rules on single-stock leveraged ETFs, with blue-chips like Micron having surged hugely, and as investors start to reassess high-bandwidth memory (HBM) supply and demand, AI capital spending sustainability, and industry cycle inflection points, capital is significantly reducing risk appetite for high-valuation semiconductor sectors.
Korean Regulatory Tightening of Leveraged ETFs Becomes Direct Trigger for This Correction
This sell-off first began in Asian markets.
On July 16, South Korea’s Financial Services Commission (FSC) officially announced it would tighten regulatory measures on single-stock leveraged ETFs, including: raising the minimum margin from 10 million KRW to 30 million KRW; only allowing cash for margin; a limit of 20 shares per single-stock leveraged trade; and a ban on new single-stock leveraged products.
After the announcement, Korean storage chip stocks were first hit hard, with SK Hynix once plunging over 11% and Samsung Electronics dropping more than 8%, with losses quickly spreading to European and U.S. markets.
According to analysis cited by Daily Economic News, Korea’s move directly targeted the most active leveraged funds in storage stock trading.
Over the past year, driven by demand for AI servers and HBM, storage leaders like Micron and SK Hynix became key targets for South Korean retail investors and leveraged ETF funds. The regulatory tightening means this inflow of funds is now restricted, and leverage reduction in these products further amplifies share price volatility.
JPMorgan analyst Nikolaos Panigirtzoglou pointed out, “Since the peak in June, storage chip leveraged ETF AUM has shrunk 34%, while all leveraged equity ETFs fell only 13% over the same period.”
He noted that what is more noteworthy is that the size of leveraged ETF assets for storage stocks is about three times that of ordinary equity ETFs relative to relevant companies’ market value—making them important volatility amplifiers in the sector. Once stocks start adjusting, the leveraged ETFs’ daily rebalancing before close further exacerbates the decline.
Three Main Market Repricing Logics: Capital Flows, Fundamentals, and Valuation
But Korean regulation is only the trigger this time.
Based on recent Wall Street discussions, the market is in fact repricing the AI storage sector around three main themes: capital flow, company fundamentals, and valuations.
- Capital flows: Deleveraging leveraged ETFs further amplifies volatility
Barron's pointed out that an important driver for the rally in storage stocks like Micron over the past year has been single-stock ETFs providing leveraged exposure.
These products maintain fixed leverage ratios through derivatives such as options and swaps. Once stock prices start to fall, they need to rebalance before daily close, constantly selling relevant assets and creating a “fall—reduce positions—further fall” negative feedback loop.
JPMorgan analyst Nikolaos Panigirtzoglou said that since the June peak, storage chip leveraged ETF AUM is down 34%, whereas all leveraged equity ETFs are down only 13% for the same period.
He pointed out that leveraged ETF assets for storage stocks are about three times that of regular equity ETFs as a proportion of the underlying market cap, making them important volatility amplifiers. This tightening of Korean regulation is further accelerating the deleveraging process.
- Fundamentals: Market is watching if supply-demand turning point is near
Aside from capital flows, the market is also reassessing the supply-demand landscape for the storage industry in the next several years.
Barron’s pointed out that Dutch lithography leader ASML recently said that the new generation of EUV lithography equipment can further improve storage chip manufacturing efficiency.
The market expects SK Hynix and Samsung Electronics to both roll out next-generation EUV equipment, meaning the future expansion speed of HBM, DRAM, and other high-end storage capacity might be faster than previously expected.
Any news that could ease tight storage supply can prompt market concerns that the “super cycle” may not last as long as previously thought.
Meanwhile, research analyst Shuli Ren recently reported that the global storage chip shortage will peak in Q2 of 2026, and begin gradually balancing in Q1 2027, with possible sector oversupply returning as early as 2028.
This means that compared to the traditional storage cycle, AI has indeed extended the industry upcycle but has not eliminated its cyclical nature.
- Valuations: Sustainability of AI capital spending comes under scrutiny
Another key sensitivity is whether AI capital expenditure can continue to grow at a high pace.
Barron’s noted that Reuters reported this week that AI cloud company CoreWeave is studying the use of financial instruments to hedge risks of future memory and storage prices falling.
Although the report did not show CoreWeave expects prices to imminently drop, nor is there any sign the company plans to cut AI capital spending, some investors saw the news as the industrial chain starting to anticipate potential future price cycles.
Meanwhile, rising geopolitical risk is prompting some capital to move from high-valuation growth stocks to defensive assets. Melius Research tech research head Ben Reitzes said that as risk appetite decreases, money is leaving the previously crowded AI trades, and storage chips—at high valuations—are now a focus of profit-taking.
For Micron, which has surged nearly 700%, after the stock price has fully priced in optimism, any news about slowing AI demand, increased supply, or price cycle changes can trigger market repricing.
Wall Street Still Bullish on HBM, But “Best Expectations” May Be Priced In
It’s worth noting that the sharp drop does not mean institutions have collectively turned bearish.
KeyBanc analyst John Vinh still maintains an optimistic view of the storage sector, forecasting:
- NAND prices to rise 30% to 40% in Q3;
- Another approximately 15% in Q4;
- HBM prices could double by 2027 compared to current levels.
Daniel Morgan, senior portfolio manager at Synovus Trust, also said that in the upcoming earnings season for tech giants, cloud providers are expected to again highlight a lack of compute power, and data center storage remains one of the biggest bottlenecks for AI servers. He expects data center storage prices to rise another 30% or so in Q3.
On the other hand, research analyst Shuli Ren recently pointed out that the current global storage chip shortage is expected to peak in Q2 2026, begin to balance out in Q1 2027, and potentially swing to oversupply by 2028.
This means that compared to traditional storage cycles, AI has indeed prolonged the industry boom, but has not completely removed cyclicality.
For the market, the real question is not whether HBM demand will remain strong, but whether current stock prices have already priced in several years of optimism.
As AI compute investments start to bear fruit, the market is focusing more on future supply-demand balance, the sustainability of capital spending, and valuation safety margin. After a huge rally in the past year, the storage chip sector is shifting from “trading only AI demand” to “trading both AI demand and cycle risk,” with significantly higher volatility as a result.
Risk Warning and DisclaimerMarkets are risky, and investment requires caution. This article does not constitute personal investment advice and does not take into account individual users’ specific investment objectives, financial situation, or needs. Users should consider whether any opinions, views, or conclusions in this article are suitable for their particular circumstances. Investment decisions based on this content are at your own risk. ```