Post

Earnings surged 372%, but stock prices fell 19% — Wall Street has finally said "No" to AI storage Looking at SanDisk's earnings report — revenue of $8.97 billion, a year-over-year surge of 372%, gross margin of 84.6%, EPS crushing expectations, plus a $14 billion buyback plan. Looking at Western Digital's earnings report — revenue of $3.75 billion, up 44% year-over-year, net profit soaring 12 times year-over-year. Then waking up — Western Digital dropped 19%, SanDisk dropped 12%, SK Hynix dropped 8%. Everyone was stunned. "Isn't the performance explosive? Why the drop?" Why? Because Wall Street now only cares about "how much can be earned in the future," not "how much was earned in the past." SanDisk's guidance midpoint is $10.55 billion, while the market expected $11.148 billion. Less than a 6% difference. Stock price crashed 12%. Western Digital is even more unfair — next quarter's revenue guidance midpoint is $4.1 billion, analysts expected only $4.02 billion. Yet it fell 11% after hours and another 13% the next day. The better the earnings, the worse the drop. This scene has repeatedly played out this year — SK Hynix's Q2 earnings caused a 30% intraday drop, Samsung's profit surged 1810% but stock price fell 6%. This is not a problem of one or two companies — it's the market's pricing logic for the entire AI storage supercycle loosening. Goldman Sachs summed up the essence in one sentence: "The core contradiction currently facing the storage industry is not a deterioration in fundamentals, but that market expectations have excessively outpaced reality." In plain language — It's not that storage is failing, but that market expectations for storage have reached a point where "you must be perfect, must beat expectations every time, and must significantly raise guidance." SanDisk has risen over 460% this year, Western Digital up 200%. The good news has long been priced in. When earnings land and new catalysts for beating expectations don't appear — profit-taking intensifies. What’s even more worrying is another signal. NVIDIA is evaluating reducing the HBM configuration of Rubin Ultra. Originally planned to use HBM4e 12hi, now considering various lower configurations like HBM4e 8hi, HBM4 12hi. Even NVIDIA can't get enough high-end HBM. What does this mean? HBM is not a money printer; it’s a bottleneck. Storage supply and demand are tight — on one hand supporting prices, on the other limiting AI chip shipments. The story of "supply not meeting demand" is starting to show another face. But the market is deeply divided. Citibank says: inventory is low, supply fulfillment rate dropped from 70% to 50%, capacity can't meet orders. China Asset Management says: in 3 to 5 years, storage supply will still struggle to meet all demand. Morgan Stanley says: memory contract prices will peak in Q4. Renqiao says: the industry has peaked, the extremely high-profit phase will inevitably be short-lived. Both bulls and bears agree earnings are strong. The disagreement is: how long can this strength last? So my judgment is simple — The plunge in storage stocks is not due to fundamentals collapsing, but "expectation gaps" killing them. In the past year, the market priced AI storage as "getting better every quarter." Now SanDisk says "gross margin is flat" — interpreted as a peak. Western Digital says "next quarter guidance midpoint $4.1 billion" — slightly above expectations, yet still falls. When the market is used to "beating expectations," anything "meeting expectations" is considered falling short. Trading advice? Three sentences: First, don’t comfort yourself with "good earnings." This earnings season, the market looks at guidance, gross margin trends, and the "second derivative" — how fast it grows is less important than whether it can grow faster. Second, volatility in the storage sector is not over. SanDisk plunged 47% in July alone, wiping out $150 billion in market value. SK Hynix hit 30% limit down in pre-market on Nextrade. When liquidity dries up, fundamentals don’t matter. Third, if you hold related assets — prepare for volatility. Who is right or wrong between bulls and bears is irrelevant; uncertainty itself is reason enough to reduce positions. Wall Street is voting with its feet to tell you — The best days of AI storage may already be behind us. Or rather, the market thinks they are behind us. $SNDK $WDC $SKHY #存储股财报后下挫，AI内存牛市还稳吗？