Bernstein Goes Contrarian Bullish on Storage Stocks: Memory ASP Slowdown Fears Overdone, Top Pick Is Seagate
nashnova research
Bernstein maintains its constructive view on memory and HDD stocks, arguing the recent pullback has over-discounted slowing price gains; Seagate remains the top pick, with FY2027 earnings estimates running ~12% above consensus for all three names.
Has memory pricing really peaked?
The bears' main argument: DRAM and NAND price gains are decelerating → the cycle has topped → a sharp downturn follows.
Bernstein pushes back point by point. The core claim: deceleration is relative to an unprecedented first-half surge — absolute pricing power remains extreme.
This means → going from +90% quarter-on-quarter to +15% sounds like a brake. But a 15% QoQ gain still sits above the 90th percentile of the past 15 years — prices have almost never been higher.
What do the "slowing" numbers actually look like?
TrendForce data: NAND contract prices rose 85–90% QoQ in Q1, slowing to an expected 15–20% in Q4. DRAM went from 93–98% in Q1 to an expected 10–15% in Q4.
In plain terms = prices are decelerating, but off a base that has already quintupled. The Q4 increase of roughly 5 cents per GB still ranks at the 95th percentile in absolute dollar terms over 15 years.
This reflects an extreme operating-leverage moment: nearly all incremental ASP drops straight to gross profit. Bernstein calls it the strongest operating leverage the industry has seen in over a decade.
How are long-term agreements "diluting" the headline price gains?
Most memory vendors have locked significant volumes into long-term agreements (LTAs) with price floors and caps: SanDisk — roughly two-thirds of bits by FY2028; Micron — over one-third of revenue locked through 2030; Kioxia — targeting half by 2028.
Bernstein models the mechanics: if spot prices hit the cap and LTA penetration is 40%, a 30% spot increase translates to only an 18% blended ASP gain. At 60% penetration, it shrinks to 12%.
This means → as more bits ship under contract pricing, headline ASP growth mechanically compresses — but this is not weakening demand. It is contract structure absorbing the gain. Bernstein argues the market is misreading this as a cycle rollover, calling it the key expectations gap and a potential catalyst for re-rating.
Why does SanDisk (SNDK) stand out?
SanDisk is one of the few pure-play NAND names globally. Its data-center bit share surged from 12% in FY2025 Q4 to 38% in FY2026 Q4.
Downside protection: an average of 53% of shipped bits in FY2027–2030 sit under LTAs, implying a floor ASP of $0.26/GB — only 19% below its latest blended ASP.
In plain terms = the LTA structure puts a floor under SanDisk's earnings — even if prices fall, the drop is capped at less than a fifth. Bernstein sees the current 5.2x FY2028 P/E as too low, because the earnings trough this cycle is structurally higher than anything before.
Are HDD stocks the better bet?
Bernstein judges HDDs are at an earlier stage of the up-cycle: nearline HDD ASP per exabyte is up only 13.6% YoY, versus ~400% for NAND and DRAM over the same period.
This means → HDDs carry none of the "decelerating gains" narrative baggage. They have room to run higher and longer.
Seagate (STX) remains Bernstein's top pick; Western Digital (WDC) and SanDisk (SNDK) are also rated outperform.
What is Bernstein really betting on?
The one-line thesis: the market is misreading LTA-cap-driven headline deceleration as demand weakness.
This reflects a valuation regime where the pricing mechanism has changed but investors are still applying the old framework to a new structure.
Bernstein's FY2027 earnings estimates for all three names sit roughly 12% above consensus — the bet is that this expectations gap will close.
市场有风险，内容仅供研究参考，不构成投资建议。