Bernstein: Communication Equipment Is the Most Crowded Trade; Sandisk May Be Mispriced
nashnova research
Bernstein's latest hardware crowding report shows communications equipment positioning at a historical 95th-percentile extreme, while Sandisk — up over 400% year-to-date — sits near the bottom of the crowding scale, flagging a potential mispricing in a pure NAND beneficiary.
What is a "crowding score" and why does it matter?
Bernstein builds a composite crowding score from five observable proxies: active-fund overweight intensity, trade persistence, long-term price momentum, sell-side buy-rating concentration, and the gap between high expectations and implied achievability.
In plain terms = it measures how many investors are already on the same side of the trade — a crowded stock doesn't have to underperform, but when drawdowns hit, crowded names tend to fall harder.
This is a risk-management tool, not a directional signal.
Which subsectors are most crowded?
AI capex redefined servers, accelerator chips, and optical interconnects as "physical constraints," funneling capital into hardware subsectors that spent the prior two decades starved of interest.
Communications equipment leads with an average Z-score of roughly 0.74; its crowded-stock share sits near the 95th percentile historically. Semiconductors score about 0.31, electronic equipment about 0.28, and computers & peripherals are near neutral.
This reflects a structural irony: subsectors known for weak long-term returns are now carrying active overweights, momentum, and consensus buy ratings all at once.
Which stocks are the most — and least — crowded?
At the crowded end: AMD (Z-score ~2.35, ~97th percentile), Arista (~2.20, 94th), Applied Materials (~1.69, 91st); Broadcom at ~1.62 is also near the top.
Nvidia sits at roughly 75%, Apple at about 59% — upper-mid range.
Within semis the gradient is clear: AI chips (Broadcom, AMD, Nvidia) most crowded → equipment (Applied Materials, Lam Research, KLA) next → analog (ADI, NXP, Texas Instruments) after that. KLA has visibly lagged since June — Bernstein attributes this to slower CY2026 growth expectations, ~100 bps of DRAM-driven gross-margin pressure, and a valuation premium still above peers.
What does Qualcomm's "narrative pulse" tell us?
Qualcomm has spent most of 2026 in the lowest decile; its crowding percentile is roughly 4%.
A brief squeeze followed data-center-related announcements in April–May — the stock roughly doubled — but the crowding score quickly faded back.
This means → a narrative pulse can manufacture short-term crowding, but without order-book and earnings visibility to sustain it, the score retreats fast. Bernstein rates Qualcomm market-perform with a $165 target.
Why might Sandisk be mispriced?
Sandisk is a pure-play NAND-upcycle beneficiary — up over 400% YTD, peaking near 900% in June — yet its crowding percentile has sat at roughly 9–10 since July as the market worries about decelerating ASP gains.
Bernstein sees structural NAND undersupply: more than half of FY2027 bits and roughly two-thirds of FY2028 bits are already sold under long-term agreements with price floors and ceilings, covering eight datacenter and edge customers, backed by ~$16.5 billion in financial guarantees.
In plain terms = a big price move is not the same as crowded positioning, and low crowding doesn't mean the score is reliable — Sandisk only re-listed in February 2025, so some factors need two years of rolling data and won't stabilize until around February 2027. Rating: outperform, target $3,000.
What risk signals surround Apple and the storage subsector?
The storage subsector (Western Digital, Seagate) is the most crowded in IT hardware. Western Digital has held a crowding percentile of 1 every month since June 2025 — the longest streak at the ceiling in Bernstein's coverage.
Apple keeps its outperform rating and $370 target, but Bernstein flags downside risk to December-quarter gross margin and EPS: DRAM and NAND costs account for roughly 90% of the year-over-year cost increase in the new device lineup.
This means → sustained ceiling-level crowding combined with margin pressure makes the stock more vulnerable to a negative catalyst if EPS misses.
市场有风险，内容仅供研究参考，不构成投资建议。