SMH Returned Over 120% in One Year. Is the VanEck Semiconductor ETF Still a Buy?

The VanEck Semiconductor ETF (SMH) has gained 120% over the past year and averaged a 51% annual return over three years, supported by AI-driven chip demand and megacap tech capex. The fund trades at about 22x forward earnings, while global semiconductor equipment sales are projected to rise from a record $133 billion in 2025 to $156 billion by 2027. The piece argues the sector still has upside, though it is more of a valuation and allocation commentary than a fresh catalyst.

Analysis

The important second-order effect is that this is no longer just a “chip designers” trade; it is a capex-transmission trade. As hyperscalers keep extending AI build plans, the beneficiaries broaden to equipment, specialty materials, test/packaging, and power-management names, which means the next leg may come from companies with lower headline AI beta but better earnings durability. That broadening usually extends the cycle because it lowers dependence on one or two mega winners and turns the theme into a multi-quarter industrial ordering story.

Valuation is not the primary risk; the risk is marginal demand disappointment versus elevated expectations. When a sector compounds this hard, even a 5-10% reset in AI capex guidance can trigger a much larger de-rating in the most crowded names because positioning is likely already extended. The concentration profile matters: the ETF’s upside remains tied to a handful of mega-cap flows, so any pause in the AI narrative would hit the basket faster than the underlying end-market would deteriorate.

The contrarian opportunity is to lean into the less obvious winners while fading the most consensus-heavy exposure. Equipment and diversified semi-supply-chain names should hold up better if AI spending stays strong but hardware revenue recognition lags, because investors will pay up for cleaner order visibility. On the other side, legacy compute franchise names with weaker AI participation can lag even in a strong tape, creating useful pair-trade structure.

Near term, the catalyst path is straightforward: capex updates over the next 1-2 quarters and equipment order commentary are the key checks. If those remain firm, the trade likely grinds higher rather than gaps higher, which favors call spreads over outright equity. If guidance rolls over, the fastest air pocket should be in the most concentrated mega-cap semiconductor exposure, not the broader supply chain.

Key Decisions for Investors

• Long AMAT / LRCX basket vs. short a concentrated semiconductor ETF sleeve for the next 2-3 quarters: benefit from AI capex transmission while reducing single-name concentration risk; target 1.5-2.0x downside capture if the cycle cools, with cleaner earnings visibility than the index.
• Buy SOXX over SMH on any 3-5% pullback: slightly lower concentration and less single-name dependency should outperform if the sector continues up but leadership narrows; use as a lower-volatility expression of the same theme.
• Sell SMH upside via 1-3 month call spreads rather than outright shorts: premium remains rich after the run, and this structure monetizes a grind-up scenario while defining risk if AI capex surprises higher again.
• Pair long NVDA against short a weaker legacy compute/semicap proxy over 1-2 quarters: stay with the name that has the most pricing power and fastest AI monetization, but hedge with a leg that is more exposed to valuation compression if growth normalizes.