This ETF Trounced the S&P 500 in 2025. Here's Why It Could Do It Again Next Year

The VanEck Semiconductor ETF (SMH) has outperformed the S&P 500, rising roughly 39.5% year-to-date through Dec. 17, 2025, driven by heavyweight constituents Nvidia, Taiwan Semiconductor Manufacturing and Broadcom, which together comprise more than a third of the fund and are reporting strong revenue growth. SMH trades at a premium (P/E ~38.6) to the Vanguard S&P 500 ETF (P/E ~28.5), but the article argues that continued AI data-center buildout and secular semiconductor demand justify the valuation while noting bubble/valuation risk.

Analysis

Market structure: The AI-driven datacenter buildout concentrates demand on a small set of premium suppliers (NVDA, TSM, AVGO, AMD, MU and equipment vendors), giving them near-term pricing power and pushing SMH P/E (~38.6) well above the S&P (28.5). Lead times for advanced nodes and HBM memory (6–18 months) create a persistent supply inelasticity that supports >10% revenue growth consensus for the top 3 names into 2026, while commodity-exposed legacy fabs and low-end IDM players face margin compression.

Risk assessment: Tail risks include an AI demand shock (>=30% slowdown vs current forecasts), tightened export controls to China, or a TSMC fab outage — each could trigger a >25–40% drawdown in top holdings. Near-term catalysts: NVDA earnings (next 30–60 days), TSMC capex guide, and US export policy updates; medium-term (3–12 months) risks center on memory cycle swings and potential fab overbuild.

Trade implications: Favor concentrated exposure to market leaders rather than broad-cap semis: trade selection should overweight NVDA (software+GPU moat) and TSM (foundry scarcity) while using options to cap drawdowns. Consider pair trades to neutralize market beta (long NVDA/TSM vs short SPY or sold-delta strategies) and use 3–9 month timeframes to ride AI adoption while protecting for mean reversion.

Contrarian angles: Consensus understates concentration and capex risk — SMH gains are dominated by three stocks (>33% weight), so beta to NVDA is extreme; historical parallel is 2003–06 capex-fueled cycles where mid-cycle overbuild created multi-year ASP declines. Unintended consequence: accelerated fab spending to shore supply could fragment global capacity (higher structural costs) and compress returns on new fabs by 10–20% over 3 years.

Key Decisions for Investors

• Establish a 3–5% portfolio long in NVDA (ticker NVDA) via either shares or a 6–9 month 10–15% OTM call spread to limit premium; take profits if NVDA rallies +40% from entry or reduce position if next two quarters' revenue guidance misses by >10%.
• Add 2–3% long in TSM (TSM) and 1–2% long in AVGO (AVGO) to capture foundry and connectivity exposure; trim both if TSMC announces capex cut >15% YoY or if TWD weakens >5% vs USD within 60 days.
• Implement a relative-value trade: long SMH (2–3%) financed by a short SPY position (equal dollar, 2–3%) to isolate semiconductor alpha; unwind if SMH underperforms SPY by >8% over a rolling 30-day window or after NVDA miss.
• Buy defensive insurance: allocate 0.5–1% of portfolio to 3-month 10% OTM puts on NVDA or 7% OTM puts on SMH (size to cover core positions) to protect against a >20% drawdown triggered by an earnings shock or regulatory action.