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

The iShares Semiconductor ETF (SOXX) rose 23% in May as chip shortages, AI-related demand, and strong earnings drove a broad semiconductor rally. Micron and AMD have become the ETF’s top holdings, with Micron benefiting from memory shortages and an UBS target hike to $1,650, while AMD beat Q1 expectations with revenue up 38% to $10.3B and adjusted net income up 45% to $2.27B. Broadcom’s softer-than-expected AI chip sales outlook triggered a 6% SOXX drop on Thursday, highlighting some near-term volatility despite the still-positive sector narrative.

Analysis

The market is now pricing semis as a scarcity trade rather than a pure AI-capex trade, which matters because scarcity winners usually travel farther and faster than end-demand winners until supply catches up. That helps memory and CPU leverage names most, but it also creates a fragile consensus: the more the group is driven by bottlenecks and narrative rotation, the more vulnerable it becomes to any single guide-down or inventory normalizing comment.

Broadcom’s softer AI-chip outlook is the first real stress test for the new leadership set. The second-order implication is not that AI demand is breaking, but that hyperscalers are increasingly multi-sourcing and optimizing for cost/performance, which compresses the moat of bespoke/custom silicon vendors faster than the market expects. That dynamic is constructive for diversified suppliers with content across memory, compute, and networking, but it also raises the odds of violent relative rotations inside the sector rather than a clean index-level trend.

The contrarian risk is that the semiconductor tape has moved ahead of revisions. If the next two earnings cycles confirm only modestly better-than-expected demand rather than sustained reacceleration, the ETF can de-rate quickly because positioning is crowded and momentum-driven. The key tell over the next 2-6 weeks is whether management teams raise not just current-quarter numbers but also capacity expansion plans and lead-time commentary; absent that, the trade shifts from scarcity premium to peak-narrative risk.

Bottom line: this is still a favorable backdrop for names with near-term estimate revisions and pricing power, but the asymmetry is better in relative-value expressions than outright beta. I would fade the broad ETF if the group fails to widen beyond the current leaders, while leaning into the clearest estimate-upgrade beneficiaries on pullbacks.

Key Decisions for Investors

• Long AMD vs. short SOXX for 4-8 weeks: AMD has the cleanest near-term revisions momentum, while the ETF is exposed to broad de-risking if Broadcom-like guidance concerns spread; target 8-12% relative outperformance, stop if SOXX reclaims leadership on strong breadth.
• Add MRVL on weakness into the next 1-2 weeks: if the market continues rewarding non-Nvidia compute/networking exposure, MRVL offers upside from content growth without requiring perfect custom-chip execution; upside is 15-20% if sector breadth improves, but cut if customers signal inventory digestion.
• Reduce or hedge AVGO into earnings-driven rallies: the stock has become the sector’s sentiment barometer, and any further evidence of customer diversification could compress the multiple quickly; consider a short-dated call spread or a partial short against stronger memory names.
• Stay long SMH/SOXX only as a tactical momentum trade, not a core position: use tight trailing stops and treat 5-7% drawdowns as regime-change signals; the risk/reward has shifted from breakout to late-cycle continuation.
• If you want a cleaner contrarian hedge, short a basket of the most crowded AI infrastructure names on any post-earnings spike and pair it with MRVL/AMD longs; the goal is to isolate estimate revisions while reducing exposure to multiple compression.