SK Hynix and Samsung Just Sent a Major Warning to Micron Investors

Memory chip pricing is still rising (SK Hynix DRAM +~30% sequentially; Samsung DRAM +>40%; NAND +~mid-50% at SK Hynix and +~high-60% at Samsung), but the pace is below Street expectations. Goldman cut its SK Hynix DRAM price improvement outlook to 19% for the current quarter from a prior 39% view, citing weaker pricing; HBM4 shipments also came in slower-than-expected despite management ramping HBM4 production in 2H. The article flags long-term customer agreements covering ~20% of Micron’s DRAM and ~one-third of NAND as a dampener on peak pricing (though a hedge), implying Micron’s next-quarter DRAM pricing could also fall short and peak earnings may be lower than previously expected.

Analysis

The market is probably underestimating how much of the memory upside is now being converted into a lower-but-longer earnings plateau rather than a clean peak. Long-term supply agreements reduce quarterly volatility, but they also cap the right tail of spot pricing; that is bearish for MU’s forward EPS power even if it makes the stock feel more “quality-like.” Second-order, the real beneficiaries are memory buyers like hyperscalers and handset/PC OEMs, while the losers are the upstream names whose multiple expansion was being justified by peak margin scarcity.

Near term, MU is the most exposed name because the next print will tell us whether pricing is merely normalizing or already rolling over faster than consensus. The key catalyst is not the current quarter but the next 1-3 months of commentary on HBM mix, contract coverage, and whether DRAM guidance gets ratcheted down again. If HBM4 ramps lag while contract penetration rises, the cycle can stay “supported” on paper while still printing lower peak returns on capital.

Over 6-18 months, the contrarian risk is that investors treat contract stability as a de-risking event when it may actually pull demand forward and lengthen the downside. If pricing re-accelerates, the thesis breaks; if MU shows sequential DRAM pricing below consensus again, multiple compression should continue. This is less a call on semiconductor demand broadly than on whether the market is paying peak-cycle multiples for a mid-cycle earnings base.

Key Decisions for Investors

• Short MU into the next earnings window or via a 1-2 month put spread; thesis is that consensus is still too high on DRAM pricing and HBM mix. Falsify if management re-accelerates DRAM guide and HBM4 volume surprises higher.
• Use MU weakness as a relative-value short against a broader AI semi basket (e.g., long SOXX / short MU) for 1-3 months; the trade isolates memory-cycle downside while dampening factor risk.
• Do not buy the post-rally dip in SSNLF or SKHY on ‘AI scarcity’ headlines until there is evidence pricing is re-accelerating, not just stabilizing; watch for a second consecutive quarter of weaker-than-expected price realization.
• Set an alert for MU guidance on contract mix and DRAM price growth: if sequential DRAM pricing comes in below the low-20% area, treat that as confirmation of a lower peak and add to the short.
• If HBM4 shipment commentary turns clearly positive before MU reports, cover short exposure immediately; that would be the fastest path to a squeeze.