Back to News
Market Impact: 0.45

Why the iShares Semiconductor ETF (SOXX) Jumped 23% in May

Technology & InnovationCorporate EarningsCorporate Guidance & OutlookCompany FundamentalsAnalyst EstimatesMarket Technicals & FlowsInvestor Sentiment & PositioningArtificial Intelligence

The iShares Semiconductor ETF (SOXX) jumped 23% in May as chip shortages, strong earnings, and AI-related demand pushed semiconductors higher. Micron and AMD emerged as the ETF’s top holdings, with Micron benefiting from memory shortages and AMD rising 45% after a strong Q1 earnings beat and better-than-expected guidance. Broadcom’s weaker-than-expected AI chip sales outlook triggered a 6% drop in SOXX on Thursday, highlighting near-term volatility despite the broader bullish sector trend.

Analysis

The rally is morphing from a single-name AI leader trade into a broader shortage/capacity trade, which usually lasts longer but becomes more fragile at the index level. That shift is important because the beneficiaries now include memory and CPU names with more cyclical earnings sensitivity, meaning upside can persist even if AI enthusiasm cools, as long as lead times and pricing remain tight. The second-order effect is that the market is implicitly bidding in future capex scarcity, not just near-term revenue beats.

Broadcom’s weaker reaction is a useful tell: when a bellwether with direct AI exposure cannot exceed elevated expectations, the semiconductor tape becomes vulnerable to factor rotation and profit-taking. The key risk over the next 1-4 weeks is not a collapse in fundamentals, but multiple compression as positioning gets crowded and investors realize the “good news” is already forward-loaded into the largest index weights. That makes the ETF more exposed to guidance variance than to headline demand headlines.

There is also a competitive spillover: if hyperscalers diversify custom silicon, the immediate loser is the concentrated AI accelerator narrative, but the broader ecosystem can still win through networking, memory, and packaging bottlenecks. The market appears to be underweighting how quickly surplus capital could be redirected from compute to memory/analog/equipment if pricing power persists for another quarter or two. Conversely, if one of the top holders disappoints on supply normalization, the unwind could be sharp because recent gains have been momentum-driven rather than purely fundamental re-rating.

The contrarian view is that the sector may be less a pristine AI trade and more a late-cycle squeeze trade. If inventories normalize faster than expected or enterprise inference spend is slower to monetize, the current leadership can flip from “scarcity premium” to “growth deceleration” almost immediately, particularly over the next earnings season.