The last time SK Hynix Inc. (NASDAQ:SKHY) suffered a monthly decline this severe, Lehman Brothers had just collapsed.
Shares of the South Korean memory giant have plunged roughly 41% this month, marking their worst monthly performance since October 2008.
The selloff accelerated Tuesday, with the stock dropping another 14% in Seoul just one day before second-quarter earnings.
SK Hynix will report second-quarter earnings on Wednesday, after the close.
Chart: SK Hynix Eyes Worst Monthly Performance Since October 2008

China Changed the Memory Narrative
The memory trade has become less about current profits and more about future supply.
The catalyst was ChangXin Memory Technologies‘ blockbuster Shanghai debut, which sent shares soaring more than 460% on the first trading day and reignited concerns that China could rapidly expand DRAM production.
That possibility triggered a broad selloff across global memory stocks, including Micron Technology Inc. (NASDAQ:MU), SanDisk Corp. (NASDAQ:SNDK), SK Hynix and Samsung Electronics Co Ltd. (OTC:SSNLF), as investors began pricing in a future normalization of memory prices rather than today’s record profitability.
Only days ago, Wall Street viewed memory as one of AI’s clearest winners.
Now the market is asking whether the industry’s pricing power is already peaking.
All Eyes On Q2 Earnings
Yet fundamentals tell almost the opposite story.
Wall Street expects SK Hynix to report second-quarter revenue of $55.7 billion and earnings per share of $4.79 on Wednesday.
That would represent a 243% year-over-year increase in revenue and a 604% surge in EPS, underscoring just how much investors expect the AI-driven memory boom to continue.
Those forecasts imply another quarter of extraordinary growth driven by high-bandwidth memory (HBM), the chips powering Nvidia’s AI accelerators.
For six consecutive quarters, SK Hynix has either met or comfortably exceeded Wall Street’s expectations.
The company beat earnings estimates by nearly 50% last quarter and revenue topped consensus by more than $2.3 billion.
Since the start of 2025, earnings surprises have repeatedly ranged between roughly 20% and 65%, while revenue has missed estimates only once—and by a negligible amount.
The market barely reacted after most of those reports, suggesting investors had already priced in the good news.
| Fiscal Quarter | Revenue (Reported) | Revenue Surprise | EPS (Reported) | EPS Surprise | 1-Day Stock Reaction |
|---|---|---|---|---|---|
| Q1 2026 | $35.55B | +$2.31B (+6.95%) | $3.83 | +$1.28 (+50.27%) | +0.16% |
| Q4 2025 | $22.95B | +$1.36B (+6.28%) | $1.45 | +$0.22 (+18.25%) | +2.38% |
| Q3 2025 | $17.08B | −$19.3M (−0.11%) | $1.25 | +$0.35 (+38.17%) | +7.10% |
| Q2 2025 | $16.18B | +$1.26B (+8.45%) | $0.68 | +$0.02 (+3.66%) | +0.19% |
| Q1 2025 | $12.37B | +$299M (+2.48%) | $0.80 | +$0.35 (+78.13%) | −1.49% |
| Q4 2024 | $13.76B | +$64M (+0.47%) | $0.81 | +$0.19 (+30.30%) | −2.66% |
The Long-Term AI Story Is Getting Bigger, Not Smaller
The irony is that SK Hynix’s sharp correction comes just as the industry’s largest customers are doubling down on AI infrastructure.
This week, Nvidia Corp. (NASDAQ:NVDA) and South Korea’s SK Group unveiled a partnership tied to more than $500 billion of planned AI infrastructure investments.
The initiative includes large-scale AI data centers, next-generation memory development and a strategic collaboration with SK Hynix on HBM.
The announcement reinforces a structural trend that investors have been following for more than two years: AI models are becoming larger, inference workloads are rising and every new generation of Nvidia chips requires more advanced memory.
That is precisely where SK Hynix sits today.
The disconnect, then, is striking. The long-term demand outlook appears stronger than ever, while the stock is trading as though the memory cycle is already ending.
That makes Wednesday’s earnings less about the quarter that just ended and more about whether management can convince investors that the AI buildout still has years—not quarters—left to run.
Photo: Samuel Bolvin / Shutterstock
Login to comment