Why Sandisk Stock Is Still Dropping

Sandisk (SNDK) shares fell a second straight day, dropping 9.6% through 11:15 a.m. ET as investors reacted to TSMC’s Q2 beat (+77% YoY profit) alongside a higher capex outlook. TSMC warned it will spend $60B+ on capital investment in 2024 vs ~$54B prior, prompting concerns about free-cash-flow pressure. Despite the sell-off, the article argues the spend supports AI-driven demand for NAND flash pairing with TSMC CPUs/GPUs, keeping Sandisk’s near-term profit outlook supported.

Analysis

The immediate selloff looks more like a margin/positioning reaction than a true demand signal. Higher foundry capex usually hits the multiple first because it pushes out free-cash-flow conversion, while the revenue benefit shows up with a lag; that creates a window where TSM can underperform even if the AI build cycle remains intact.

The second-order winner is the broader AI supply chain, but not uniformly. Compute-heavy names and equipment vendors are better positioned than NAND, and within memory the cleaner lever is usually DRAM/HBM rather than NAND storage. SNDK can still work if AI server deployments broaden storage attach rates, but the market is likely overstating how directly AI capex flows into NAND pricing; that link is weaker and more cyclical.

Contrarian risk: consensus may be chasing the phrase "AI memory" without separating compute from storage. If the current capex wave is front-loaded into leading-edge logic and HBM, NAND could lag even as the semis complex rallies, leaving SNDK as a low-quality beneficiary. Watch for any sign of memory ASP deceleration or capex discipline in the next 1-2 earnings cycles; that would be the first falsifier for the bullish read-through.

Key Decisions for Investors

• Prefer a relative-value long NVDA / short SNDK pair for 1-3 months: NVDA has the cleaner AI monetization path, while SNDK is more exposed to a lagging, cyclical NAND tape-out. Risk/reward improves if memory pricing stays flat while AI capex keeps rising.
• Do not chase SNDK strength immediately; wait for a 2-3 day stabilization or confirmation from NAND ASP data before adding. If NAND pricing rolls over, the current AI read-through likely fades quickly.
• Consider short TSM on rallies or via put spreads over the next 2-6 weeks if the market continues to punish capex-heavy stories. Thesis fails if TSM maintains margin guidance or demonstrates faster-than-expected FCF conversion from the higher spend.
• Use SOXX or SMH as the cleaner way to express continued AI capex, rather than making SNDK the primary long. The upside is broader and less dependent on the NAND cycle, which can reverse within 1-2 quarters.