The memory sector's rally is not dead; it appears to be pausing. Micron has fallen -17% over the past month and SK Hynix’s new U.S. ADR plunged -11.5% in its first week of trading. Despite those abrupt moves, both stocks still sit massively higher over the past year - Micron up +625% and SK Hynix up +522%.
What looks like a dramatic short-term selloff is better read as an extreme deceleration in momentum rather than a structural breakdown in the market.
Scoreboard first
| Name | Price | 1D | 1W | 1M | 6M | 1Y |
|---|---|---|---|---|---|---|
| Micron (MU) | $844.57 | -6.6% | -14.8% | -17.2% | +133% | +626% |
| SK Hynix ADR (SKHY) | $156.21 | -11.5% | — | — | — | — |
| SK Hynix Seoul (000660) | 0140,000 | -11.5% | -15.7% | -22.7% | +144% | +522% |
On a monthly basis the drawdowns are eye-catching, but when viewed over the 52-week band for MU - from $103 to $1,255 - the current $844 level reads more like a pullback inside a larger uptrend than a capitulation.
What the charts are saying
The technical picture displays a clear time-frame split.
- Daily timeframe - Micron registers a Strong Sell across the daily indicators. RSI sits at 41, MACD is negative, StochRSI is at zero indicating oversold conditions, and price is trading beneath the 5-day and 10-day simple moving averages.
- Weekly timeframe - The weekly read remains a Strong Buy. RSI is at 59, MACD shows +177, and ADX reads 52.9, signaling a strong trend that is still intact on the longer view.
SK Hynix’s ADR technicals are muddied by a very short trading history. An RSI value of 100 has appeared as a data artifact from that limited sample, while intraday signals have been uniformly bearish since debut. The combined interpretation: traders are harvesting profits from recent parabolic gains, but longer-cycle momentum has not yet been extinguished.
The structural bull case
The bullish thesis supporting this market move is multi-year in scope rather than tied to a single quarter.
- HBM4 pricing dynamics: estimates indicate HBM4 prices could double to $400/Gb by 2027, up from roughly $2 in H2 2026, driven by roughly three times the wafer consumption compared with standard DDR5 and prolonged yield ramp timelines measured in months.
- SK Hynix leadership outlook: the company’s CEO cautioned that the memory chip shortage may linger beyond 2030, an unusually forward-looking bullish projection from an industry leader.
- Micron's customer traction: Micron has secured 16 strategic customer agreements, including a long-term contract with Ford, broadening HBM demand beyond hyperscalers into other end markets.
- Contracting trend: by 2027, about 50% of global DRAM capacity is expected to be allocated under long-term agreements with tier-1 AI buyers, which would reduce available supply for smaller purchasers and support pricing power for contracted suppliers.
Bear case that matters
The downside scenarios are real and were catalogued during recent investor outreach.
- Chinese competition - CXMT has expanded quietly to become the world’s fourth-largest DRAM maker, and Apple is reportedly evaluating their chips for devices sold in China, a development that raises competitive risk.
- Hyperscaler demand risk - Moves such as Meta’s entry into cloud server operations could presage a slowdown in hyperscaler capital spending, shifting the market’s risk profile from supply to demand.
- Diminishing catalysts - After moves on the order of +600%, the most accessible gains may be behind investors; future appreciation will likely need to be driven by earnings and longer-term fundamentals.
- ADR debut premium erosion - SK Hynix’s ADR began trading at a premium to its Seoul-listed shares and has already experienced sharp premium compression, consistent with a post-IPO hype unwind.
The verdict
The evidence points toward a breather rather than a systemic breakdown. The underlying AI-driven memory supercycle - including HBM pricing leverage, multi-year supply tightness, and multi-year customer commitments - remains in place. What changed is that the market price accelerated too quickly, pushing daily technicals into overbought territory and prompting a rapid correction as participants took profits.
The principal risk to monitor is CXMT. If Chinese DRAM capacity scales faster than current expectations and Western hyperscalers begin qualifying CXMT as a supplier, that is the scenario in which this pause could evolve into something more serious.