History Says This Chip ETF Keeps Beating the S&P 500. It Has Trailed the Index Only Twice Since 2016.

VanEck Semiconductor ETF (SMH) has returned about 59% in 2026 versus 12% for the S&P 500 and outperformed the index in eight of the 10 calendar years through 2025. However, its concentrated semiconductor exposure produced losses roughly double the broader market in 2018 (-9% vs. -5%) and 2022 (about -33% vs. -18%), while the fund remains 15% below its June 52-week high. Nvidia represents roughly 23% of the $72 billion ETF and the top five holdings account for about half of assets, underscoring both AI-driven upside and concentration risk.

Analysis

The relevant exposure is not “semiconductors” broadly but a highly correlated AI-capex factor dominated by NVDA and, indirectly, TSM. That creates a nonlinear downside: a pause in hyperscaler GPU procurement would hit NVDA’s revenue expectations first, then transmit to TSM utilization, networking, memory, equipment orders, and ETF flows simultaneously. In a de-rating, diversification across chip subsectors offers less protection than headline holdings imply because the common variable is AI infrastructure spending.

The near-term setup is more fragile than the long-run AI thesis. After a sharp year-to-date advance and a meaningful retreat from the high, passive and momentum holders are likely to treat incremental earnings-guide revisions as the next directional signal; upside requires estimates to keep rising, while merely solid results can compress multiples over days to weeks. The key falsifier for a cautious view is another broad-based round of upward 2027 capex and semiconductor revenue revisions, led by hyperscalers and TSM’s advanced-node outlook.

A non-obvious relative winner in a moderate AI spending slowdown could be TSM: its manufacturing leadership and customer diversification make it less exposed than NVDA to a single-product digestion cycle, although it remains vulnerable to a broad capex reset and geopolitical risk. Conversely, equipment names such as ASML, KLAC, and LRCX may lag initially if customers stretch capacity additions, even if their longer-cycle demand remains intact. The consensus error is treating prior sector drawdowns as automatic buying opportunities; the recovery path depends on whether demand normalization occurs against already elevated utilization and valuation assumptions.

Key Decisions for Investors

• Do not add directional SMH exposure solely on historical return persistence; maintain any semiconductor beta below benchmark-risk limits until the next NVDA, TSM, and hyperscaler capex guidance cycle resolves over the next 1-3 months.
• For existing SMH longs, buy 3-6 month SMH put spreads financed selectively with upside call overwrites after sharp rallies. The hedge is justified if consensus forward EPS stops rising; close it if NVDA/TSM guidance drives renewed broad estimate upgrades.
• Prefer a relative-value expression of long TSM / short NVDA in equal dollar amounts for a 3-6 month horizon if NVDA’s forward estimates flatten while TSM maintains advanced-node utilization commentary. Risk: NVDA retains pricing power and accelerates earnings revisions, in which case cover the short on a renewed estimate-revision inflection.
• Set an earnings-monitor alert for hyperscaler capex guidance, NVDA data-center inventory commentary, and TSM advanced-packaging capacity updates. A synchronized cut to capex plans or evidence of order deferrals is the trigger to underweight SMH versus SPY; absent that evidence, this remains a watch item rather than a new outright short.