Why The iShares Semiconductor ETF Soared 112.8% In The First Half of 2026

The iShares Semiconductor ETF (SOXX) is up 112.8% in 1H 2026 and nearly 300% over 5 years, fueled by AI “picks and shovels” demand and memory-chip shortages. Memory-related winners are highlighted, including Micron up 314% YTD (as of July 10, 2026), while Intel is positioned to benefit from increased AI CPU demand. Despite the strong momentum, the article flags semiconductors as highly cyclical and advises against “piling in” after the >100% YTD run.

Analysis

The key market mechanism is not “AI is good for semis” but that the current leadership is being driven by a narrow, highly cyclical memory upcycle layered on top of an already crowded AI trade. That is usually the late-stage setup where earnings can stay strong while multiples peak, because investors pay for sustained scarcity right as supply response begins to compound. MU has the cleanest near-term operating leverage, but it is also the name most exposed to a reversal in spot pricing once new wafer starts and inventory normalization show up over the next 1-3 quarters.

SOXX is the vulnerable wrapper trade here: it owns the right secular exposure, but index composition dilutes the best fundamental winners and leaves it most exposed to valuation compression if AI capex growth slows even modestly. NVDA is less directly tied to memory shortages than MU, so it can keep outperforming even if the memory cycle cools, but that also means the ETF can underperform NVDA on the way up and then fall harder on the way down if investors rotate from “AI infrastructure” to “cash-flow durability.” INTC’s upside is more policy- and sentiment-driven than purely earnings-driven; that makes it a lower-quality beneficiary if the market starts demanding proof rather than narrative.

Contrarian view: the consensus is likely underestimating how quickly a shortage narrative can flip into an inventory-clearing narrative in semis. The first falsifier is not a macro recession; it is evidence that memory pricing is flattening and forward guidance is no longer being raised. If that happens, the move in SOXX can unwind faster than fundamentals, because positioning and passive flows amplify both directions over days to weeks, while the actual supply-demand correction plays out over months.

Key Decisions for Investors

• Short SOXX on strength for a 1-3 month horizon; target a 10-15% giveback if memory pricing or guidance moderates, with a tight stop if the ETF holds above recent breakout levels after the next earnings round.
• Pair trade: long MU / short SOXX for the next 4-8 weeks if you want to express the shortage theme with better alpha capture; MU has the cleanest earnings torque while SOXX is diluted by names with weaker forward revision potential.
• Underweight INTC vs NVDA in AI infrastructure baskets over the next 1-3 months; INTC’s catalyst is more narrative/policy-based, while NVDA retains the stronger secular demand engine and pricing power.
• Use put spreads on SOXX into any post-earnings pop; the risk/reward improves when implied volatility is elevated and the sector is already extended, with the thesis invalidated only if forward guidance keeps accelerating across the group.
• Set a watch item on DRAM/HBM pricing and inventory days for MU and peers; if spot prices flatten for 2 consecutive prints, expect the semiconductor complex to de-rate before reported earnings fully reflect the turn.