This month in Korea, if you’re neither an SK Hynix employee nor an owner of SK Hynix stock, you’re probably one of the unlucky ones.
After the quarterly report revealed massive profits, investment banks eager for a spectacle not only raised their profit forecasts for SK Hynix this year but also heightened employees' expectations for year-end bonuses. Using the annual practice of allocating 10% of operating profit as a bonus pool, they calculated that each employee could receive a year-end payout of several million RMB, while simultaneously casting Samsung’s executives in a harsh, uncaring light.
Afterward, anything associated with Hynix IP was met with fervent enthusiasm.
Hynix employees' uniforms have become a priority pass in South Korea's dating market; real estate agents in Anyang, where Hynix is headquartered, experienced a dreamlike quarter, with property volumes and prices surging along multiple Hynix shuttle bus routes; even the marginal China-South Korea semiconductor ETF has been pushed to a 30% premium, frequently triggering temporary trading halts.
Even the Hong Kong stock market, long criticized for lacking technological sophistication, has stepped up.
As of May 13, 2026, the Southern DWS SK Hynix Daily 2x Leveraged ETF (07709.HK), hereinafter referred to as the 2x Long Hynix ETF, listed on the Hong Kong Exchange, has reached nearly HK$60 billion in assets under management, surpassing the long-leading Tesla 2x Long ETF (TSLL.NASDAQ) on the U.S. market to become the largest single-stock leveraged derivative product globally.
No matter how niche an investment may be, when the price rises this much, just browsing the internet—even scrolling through updates from tech and gadget influencers—inevitably leads you to see enthusiastic netizens commenting directly: “Why aren’t you buying 2x long Hynix?”
Deadly leverage
On October 16, 2025, when the 2x Long SK Hynix ETF was first listed on the Hong Kong Stock Exchange, its issuance size was less than HK$5 billion. By the closing price on May 13, 2026, over seven months, the net asset value of this leveraged ETF increased by 1,011.58%, while its scale surged more than thirteenfold.
Yunjì Technology, dubbed the "first robot hotel stock," listed on the Hong Kong stock market on the same day, has seen its stock price rise sharply, yet its market capitalization remains less than four times its listing value.
You have to admit this is the terrifying efficiency of 2x leverage: SK Hynix’s underlying stock listed in Korea rose a mere 324.49% from October 17 last year to May 13 this year, yet with the momentum of a one-sided bull market, this leveraged ETF delivered an excess return of 362% beyond its theoretical 2x gain—making “3x leverage” seem a more fitting description for this explosive performance.
But if we look back over the past seven months, this paper gain is temporary.
Just two months ago, the Strait of Hormuz fell into a Schrödinger-like blockade, sending global markets into panic over sudden disruptions in oil and gas supplies. Amid the volatile and rapidly shifting situation, markets did not experience a traditional one-sided decline but instead became schizophrenic amid this atypical geopolitical conflict.
During the day, traders may still be following the safe-haven logic of “third war breaks out, supply chains disrupted,” but by night, they could suddenly shift to a short-squeeze frenzy driven by ambiguous statements from the White House spokesperson, embracing the narrative of “de-escalation and return to tech.” This ambiguity and uncertainty in the evolution of market sentiment are continuously amplified by rapid social media dissemination, translating into the capital markets as sharp sell-offs targeting tech stocks—or frantic buying on pullbacks.
Although common sense tells us that war will eventually end, and daily token consumption in the AI industry continues to accelerate, when market fluctuations become too extreme, the complexity of the process cannot be entirely ignored.
More people also experienced the volatility decay of this leveraged ETF product at this time.
Based on actual trading data from March to April 2026, Hynix's stock price declined amid severe volatility. The decline itself was problematic, but the multiple violent rebounds exceeding 10% along the way only made matters worse.
For a daily-rebalanced 2x long Hynix ETF, a one-sided decline may be tolerable, but high-volatility sideways downward movement is the real meat grinder—during its most grueling periods, it fell over 50% more than the underlying stock.
Without considering other trading fees or management fees, the product’s daily rebalancing mechanism means that in a one-sided uptrend, yesterday’s profits automatically become today’s “principal,” allowing for an additional two times leverage and generating greater excess positive returns. Conversely, in a one-sided downtrend, since the daily calculation base shrinks, the actual loss will be less than the theoretical two-fold loss.
However, once entering a sideways market characterized by alternating gains and losses, leveraged ETFs reveal their dangerous side.
The 2x Long Hynix ETF repeatedly suffers from "long-short double kills"—after rallying yesterday and rebalancing, it loses more value today during the drop, then rebalances again, only to suffer another loss tomorrow due to the reduced base when it rebounds.
Alternating gains and losses create repeated friction, causing the product's actual net asset value to decline by more than twice the extent of the underlying stock's drop, resulting in significant negative volatility drag that erodes investors' principal.
However, the market has now returned to the AI theme, with fervent capital once again converging and driving a one-sided surge that brings widespread joy.
As Hynix's market capitalization repeatedly hits new highs and leveraged ETF products with hundreds of billions in assets spark frenzied trading, the market inevitably returns to the daily question: Is this industrial revolution truly without cycles?
Silicon-based cyclical stocks
It must be acknowledged that, in terms of its listing timing, the 2x Long Hynix ETF has been blessed with the full alignment of fate, fortune, and feng shui.
For a long time prior, storage was not the absolute focus of long positions in AI on the secondary market. After all, since humanity boarded the information age train in the 1990s, storage has often been the place where euphoria is followed by carnage, with the terror of cycles far outweighing the fantasy of growth.
Memory chips (particularly traditional DRAM and NAND) are highly standardized commodities. Memory modules from different manufacturers exhibit virtually no difference in physical performance, aside from brand labels—essentially the pork stocks of the silicon world. The entire industry has long been trapped in a brutal cyclical pattern:
Shortage leads to price hikes → Major players rapidly expand production → Overcapacity → Price crash → Losses lead to reduced production → Shortage occurs again.
Each upward movement is labeled a "super cycle" amid super-optimistic expectations. Each downward movement leaves behind a trail of corpses from brutal price wars and losses in the billions.
After enduring the most brutal semiconductor winter from 2022 to 2023, the three surviving memory giants—Micron, Samsung, and SK Hynix—quietly cut capital expenditures and stopped engaging in aggressive capacity expansion that harms both competitors and themselves.
Image source: IC Insights
Then came the AI narrative, replaying the shortage-driven price hikes and essentially handing everyone a printing press.
Especially since the second half of last year, the competitive focus in the AI industry shifted from "training" to "inference," causing the primary infrastructure demand to move from "computing power" to "storage capacity," with supply bottlenecks transitioning from bandwidth to capacity, making widespread storage shortages the most popular trading narrative.
At this point, anyone who still brings up, “Wasn’t it supposed to be that AI’s end goal is electricity?” is probably missing the mark entirely.
After the third quarter of 2025, news in the AI industry has almost exclusively revolved around shortages of memory chips—first, major companies announced that HBM orders have been booked through 2027 and beyond; then, customers were notified that DDR5 is also becoming scarce, so unfortunately, regardless of whether it’s high-end or low-end, all products are now subject to price increases.
As a primary supplier of NVIDIA's HBM, SK Hynix has reaped tremendous first-mover advantages and market share, and the 2x leveraged ETF tracking SK Hynix nearly无缝 (seamlessly) rode the wave as DRAM prices soared to surpass gold, with a single box worth enough to buy an apartment in Shanghai.
So, can hopping on the AI bandwagon help you escape the pull of market cycles? What matters less is drawing conclusions now, and more is identifying where the changes will occur.
Under the HBM yield barrier, SK Hynix has established a dominant market position; in the first quarter of 2026, SK Hynix achieved a historical peak gross margin of approximately 79%, surpassing NVIDIA’s profitability during the same period.
Human nature tells us that extreme excess profits will inevitably trigger a surge in demand for capacity expansion. The apparent consensus among storage giants to "cut production" is not trustworthy in the face of absolute windfall profits.
Therefore, whether Samsung or Micron achieves a breakthrough in yield at some point in the future, undermining the narrative of HBM scarcity and widening the divergence between bulls and bears—leading to volatility in the sector—is a variable that requires ongoing monitoring.
In addition to changes on the supply side, disputes on the demand side have not been completely resolved despite the accelerating adoption of agents and increased token consumption.
Ultimately, Hynix's frenzy is built on NVIDIA's frenzy; and NVIDIA's frenzy is built on the tens of billions of dollars in annual AI capital expenditures by major downstream companies.
The marginal change in Capex remains the strongest gravitational force in the secondary market regarding all anxieties and pride surrounding AI.
Epilogue
Whether you buy or don't buy the 2x Long Hynix ETF, it will become a subtle footnote in our future reflection on this period of history.
In this era, being long and short often refers to two different things: long on faith in the AI industry, short on macro geopolitical concerns.
People always instinctively turn to history books, seeking parallels in the internet boom of the millennium or even earlier macro-level upheavals. But each technological revolution unfolds differently—and this time, the difference lies in the unprecedented speed of disruption brought by the industrial revolution.
AI is reshaping global productivity and production relations at an unprecedented pace. This extreme speed has disrupted the lengthy渗透 and fermentation processes typical of traditional technology cycles. It doesn’t grant markets time to gradually absorb valuations, nor does it offer “old guard” players a chance to benefit from the flood of liquidity through cyclical rotation.
Whether industry giants or secondary market capital, all are forced to take sides and set valuations within an extremely narrow time window, so the unit of stock price appreciation has become multiples; as a result, seasoned AI professionals now assume that six months counts as long-term in this era.
Yet, the storm over the Strait of Hormuz once again placed this wave of technological revolution under the common denominator of all previous tech cycles: industry determines the outcome and returns, while macroeconomics influences the path and volatility—the massive negative deviation of the leveraged Hynix ETF was not caused by an interruption in AI progress, but by extreme fluctuations in global macroeconomic expectations over that month-plus period.
And the real world's vulnerabilities are not limited to just the 33 kilometers at the narrowest point of the Strait of Hormuz.