This Hedge Fund Just Sold Sandisk and Bought Taiwan Semiconductor Manufacturing Stock. Does Wall Street Agree With the Move?

The article compares 1-year Wall Street upside for Taiwan Semiconductor (TSM) and SanDisk (SNDK): analysts see ~$555/share for TSM (~33% gain) versus ~$2,126/share for SNDK (~41% upside). It argues the “AI chip” demand backdrop supports both, but highlights SNDK’s memory-cycle risk as higher, implying TSM is the better risk-adjusted buy despite Sandisk’s higher upside. It also cites Druckenmiller’s Q2 move of selling SNDK and buying TSM, noting SNDK is down sharply since late June even though it remains up >500% YTD.

Analysis

TSM is the cleaner expression of AI infrastructure spend because it monetizes broad-based logic demand with better visibility, less end-market concentration, and a structurally higher quality of earnings. SNDK can keep running on scarcity narratives, but its upside is more dependent on NAND pricing staying tight; that invites a supply response from MU, WDC, Samsung, and SK hynix that can compress margins faster than the market expects once everyone chases the same “AI storage” theme.

The key second-order effect is that a strong SNDK tape can be self-defeating for the group: higher prices encourage capex, and the memory cycle tends to overshoot before reversing. TSM, by contrast, benefits from AI capex durability even if individual chip names rotate; it sits closer to the tollbooth than the cyclical commodity layer. Over 1-3 months, watch earnings commentary on inventory, lead times, and node utilization; over 6-18 months, the spread should favor TSM if AI spending broadens beyond a few hyperscalers.

The contrarian view is that the market may be underestimating how much of SNDK’s rerating already reflects the best-case scarcity story, while TSM still trades like a large-cap compounder rather than the core enabler of AI compute. What would falsify the thesis: TSM guiding to weaker advanced-node utilization or softer 2027 capex visibility, versus SNDK maintaining elevated spot pricing and positive guidance through the next two earnings cycles. If memory ASPs roll over before TSM growth reaccelerates, the relative trade likely works quickly; if not, the pair is too early.

Key Decisions for Investors

• Put on a long TSM / short SNDK pair trade on any post-earnings strength in SNDK; target 3-6 months for mean reversion in the valuation gap, with the short leg tied to memory ASP normalization and the long leg tied to AI capex durability.
• If borrow is manageable, use a 1:1 dollar-neutral pair; if borrow is tight, keep the trade cleaner by going long TSM and buying SNDK put spreads instead of an outright short.
• For tactical exposure, buy TSM on dips after guidance or macro-driven selloffs; the risk/reward is better because downside is supported by diversified demand, while upside compounds if AI node demand broadens.
• Avoid chasing SNDK after sharp rallies unless spot NAND pricing and channel inventory remain tight for another quarter; the key watch item is whether competitors expand supply, which would cap upside fast.
• Use SOXX/SMH relative strength as confirmation: if TSM leads while memory names lag, that supports staying long TSM and fading SNDK momentum over the next 1-3 months.