Why Sandisk Stock Is Still Dropping

Sandisk shares fell a second straight day, dropping 9.6% through 11:15 a.m. ET, despite Taiwan Semiconductor (TSMC) beating Q2 profit estimates with profits up 77% year over year. The sell-off appears driven by TSMC’s capex guidance rising to $60B+ in 2024 (vs ~$54B prior forecast), which investors worry could pressure free cash flow. The article argues the higher TSMC spending is actually positive for Sandisk because AI GPUs/CPUs produced by TSMC require NAND flash, keeping near-term memory demand and Sandisk profits supported even as the industry remains cyclical.

Analysis

The market is reading TSMC’s higher capex as a free-cash-flow problem, but for SNDK the more important signal is that AI system build-outs are still accelerating. The relevant mechanism is memory content per AI server: even if accelerators are the headline, incremental storage and data-movement demand can lift NAND attach rates faster than unit growth alone. That makes the immediate selloff in SNDK look more like a tape-driven de-risk than a fundamental repricing.

The second-order winner is SNDK relative to NAND peers if the industry stays disciplined on supply, because a capex-heavy leader tends to pull forward demand without instantly flooding the market. The losers are foundry names and the broader AI semiconductor basket if investors anchor on lower near-term FCF multiples rather than demand acceleration; TSM can still be right operationally and wrong in the stock if capex expands faster than earnings. Over 6-18 months, the real risk is the usual memory cycle: if capacity additions from multiple players land together, NAND pricing can roll over quickly and compress margins.

Contrarian takeaway: consensus is treating capex as supply inflation, but in AI the bottleneck is still system integration, not just wafer starts. The thesis breaks if NAND spot/contract pricing stops firming, if SNDK gross-margin guidance fails to expand next quarter, or if hyperscaler capex rolls over. Near term, this is a 1-3 month trade on pricing and sentiment; structurally, it is only bullish while supply growth remains slower than AI storage demand.

Key Decisions for Investors

• Buy SNDK on weakness over the next 1-2 sessions, targeting a 4-8 week rebound into the next memory-pricing update; use a close below the post-earnings low as the stop, because a break there would imply the market is discounting NAND ASP weakness.
• Pair trade: long SNDK / short TSM for a 1-3 month window to isolate the demand-vs-FCF misread; this works if investors keep punishing capex while AI memory demand stays intact, but cover if TSM commentary shifts from expansion to utilization weakness.