The AI boom cannot run without advanced chips. SMH brings investors closer to the semiconductor supply chain—but is its NVIDIA exposure an advantage or a concentration risk versus SOXX?
Research data cutoff: November 16, 2026, latest available U.S. session.

⚡ 10-SECOND INVESTMENT BRIEF
BOTTOM LINE: SMH gives a more direct semiconductor supply-chain exposure than thematic AI software funds or broad technology ETFs, with an especially meaningful company-concentration trade-off.
VanEck Semiconductor ETF (SMH) deserves attention for a specific reason: A more targeted semiconductor allocation for investors who believe the next stage of AI growth will require accelerators, memory, fabrication and manufacturing equipment. This is Dayhook Investing’s editorial thesis, not a promise of returns or personalized financial advice. The practical question is whether the fund’s holdings, index rules, expenses and risks actually match the investor’s goal.
Our view is generally constructive on participating in long-term AI-driven technology growth. Yet choosing the right instrument matters just as much as choosing the right theme. We compare SMH against SOXX, IYW and AIQ, explore fund-level constraints and identify the conditions that would change our conviction.
🔎 What is the SMH ETF and how does it invest?
VanEck Semiconductor ETF is managed by VanEck, was launched on December 20, 2011 and references the MVIS US Listed Semiconductor 25 Index. Its stated annual fund expense ratio in the cited documents is 0.35%. Fund fees are internal expenses deducted from returns, while commissions, bid–ask spread, tracking and individual tax treatment may add to costs.
Start with the official issuer fund page, then check its current portfolio and prospectus before committing capital. A fund’s name and the securities shown in promotional graphics can hide meaningful nuances about the benchmark and the period used for reported weights.
🏢 SMH ETF holdings list: concentration and sources
The following table reflects October 31, 2026 official VanEck fact sheet. These weights are a dated snapshot, not live positions. Rebalancing and price movements change the rankings. The weights may be different at the time a reader opens this article. The list is not a guaranteed set of future holdings or an instruction to buy the underlying equities.
| Company / symbol | Weight |
|---|---|
| NVIDIA (NVDA) | 20.19% |
| Taiwan Semiconductor (TSM) | 9.58% |
| Texas Instruments (TXN) | 5.70% |
| Broadcom (AVGO) | 5.61% |
| AMD (AMD) | 5.34% |
| Analog Devices (ADI) | 4.96% |
| Micron Technology (MU) | 4.75% |
| ASML (ASML) | 4.50% |
| KLA (KLAC) | 4.28% |
| SK hynix ADR (SKHY) | 4.14% |
Holdings/source: dated portfolio or benchmark source. Important: company weights, security listings and tracked index are separate concepts; do not confuse an index constituent list with leveraged ETF swap/collateral positions.

🔬 What Dayhook Adds: the portfolio exposure most investors overlook
SMH is best understood as the physical-infrastructure part of the AI trade, rather than a diversified technology ETF. A software breakthrough can create demand for GPUs, networking chips, HBM, advanced packaging and industrial equipment, but profitability across the chain is uneven.
The official October 31, 2026 fact sheet reports 26 holdings and a 0.35% expense ratio. Its top ten account for 69.05%, including NVIDIA at 20.19%, Taiwan Semiconductor at 9.58% and SK hynix ADR exposure. This is a deliberately narrow industry portfolio, not a defensively diversified market basket.
Dayhook is constructively biased on the structural compute demand that AI can create, but we would evaluate SMH as a cyclical satellite allocation. It is particularly well suited to investors who understand chip-cycle drawdowns, watch supply-capacity signals and can handle periods when semiconductor equities fall even as the long-term AI story remains intact.
📈 SMH ETF charts: 5 years, 1 year and 1 month
Source and timing discipline: the data cutoff for the written analysis is Nov 16, 2026 while the U.S. trading day is in progress; the following TradingView embeds update independently after publication. The charts are genuine third-party market series, not hand-drawn simulated history. Compare split-adjusted history and distributions using the fund issuer’s dated NAV total-return tables, not a picture of an intraday quote.
The October 31, 2026 VanEck fact sheet reported SMH NAV total return of +44.47% over the previous year but -14.07% for the previous month. Its five-year annualized NAV total return was +31.53% as of that month-end. The one-month figure is not annualized; performance changes with subsequent prices. Source: issuer’s dated historical performance document.
SMH — five-year market-price chart
Interactive TradingView chart; may update after publication. Historical prices and split adjustments depend on the data provider. Historical quotes.
SMH — one-year market-price chart
Interactive TradingView chart; may update after publication. Historical prices and split adjustments depend on the data provider. Historical quotes.
SMH — one-month market-price chart
Interactive TradingView chart; may update after publication. Historical prices and split adjustments depend on the data provider. Historical quotes.
⚖️ Why We Like the Exposure — and What Could Go Wrong
✅ WHY: AI computing requires physical silicon
More model inference and training generally require compute, memory, connectivity, chip packaging and equipment. SMH owns multiple segments of that infrastructure instead of only software narratives.
Exposure includes U.S. designers and suppliers, Taiwan’s major foundry ecosystem, Europe’s equipment technology and Korea’s advanced-memory firms. That is industry diversification—but also cross-border fragility.
Certain chip and equipment suppliers can book orders while enterprise AI software revenues are still developing. The thesis depends on capacity economics and persistent spending, not only enthusiasm for AI.
⚠️ RISK: Chip-cycle reversals can be severe
Chip supply can outrun demand. Memory prices, inventories and capex reductions can rapidly reverse earnings and ETF prices after strong booms.
VanEck’s Oct 31 report lists NVIDIA at 20.19%, while top ten positions totaled 69.05%. One company’s earnings and valuation matter materially.
Foundry concentration, US-China chip restrictions and global trade policy can affect demand or supply. Listing in the U.S. does not mean the business is insulated from overseas shocks.
🧭 SMH versus SOXX: the choice for different investors

What makes SMH different? SMH gives a more direct semiconductor supply-chain exposure than thematic AI software funds or broad technology ETFs, with an especially meaningful company-concentration trade-off. Comparable funds may use different indexes, weight caps, eligibility criteria and turnover rules; even when their top holdings overlap, realized returns can diverge. Consider how a new ETF would overlap with the reader’s existing AIQ, QQQ, FTEC, semiconductor and single-stock positions.
The closest ETF shown below is SOXX, but close does not mean identical. Compare the current official holdings and expense ratios of BOTH funds before drawing a conclusion from a short chart period. In particular, a two-fund comparison is more useful when it answers which risk a person wants than which fund happened to outperform last month.
SOXX — one-year comparison price chart
Interactive TradingView chart; may update after publication. Historical prices and split adjustments depend on the data provider. Historical quotes.
SOXX — one-month comparison price chart
Interactive TradingView chart; may update after publication. Historical prices and split adjustments depend on the data provider. Historical quotes.
🔭 2027 outlook: bull, base and bear scenarios
Dayhook does not invent an ETF analyst-consensus target when a verified consensus is unavailable. Rather than promising a specific next-year price, we use an explicit conditional framework.
AI infrastructure spending remains resilient, advanced packaging and memory bottlenecks are monetized, and chip suppliers retain pricing power.
AI compute grows while consumer electronics and industrial demand are mixed; winners and laggards rotate within the industry.
Inventory build, export controls or weaker hyperscaler capital expenditure compress earnings and valuations across the semiconductor chain.
Our investment judgment: A high-conviction AI chip-cycle satellite holding—not an all-weather diversified ETF. The preferred allocation size and holding period depend on the investor’s cash needs, existing technology exposure and ability to tolerate major losses. Long-term conviction in AI is not evidence that today’s quote is a guaranteed bargain.
❓ SMH ETF investor questions, risks and key facts
What are the SMH ETF top holdings?
As of October 31, VanEck listed NVIDIA, Taiwan Semiconductor, Texas Instruments, Broadcom, AMD, Analog Devices and Micron at the front of the portfolio. Weights can change.
Is SMH a good ETF for the AI semiconductor cycle?
It is a focused research candidate for readers confident in sustained AI chip demand and comfortable with semiconductor drawdowns. It is not a substitute for a diversified all-market equity fund.
SOXX vs SMH ETF: what is different?
Both target semiconductors, but SMH tracks the MVIS U.S.-listed Semiconductor 25 Index and had a high NVIDIA weight. SOXX follows a different index and constituent weighting scheme, so compare current holdings and overlap before buying both.
What is the SMH ETF price outlook for 2027?
A credible outlook depends on hyperscaler capex, AI accelerator supply, memory pricing, foundry margins, China restrictions and market valuation. The scenario table sets conditions, not a guaranteed numeric target.
Does SMH pay dividends?
It has distributions and a published 30-day SEC yield; the issuer’s Oct 31 factsheet showed 0.17% SEC yield. This is not a guaranteed distribution and not the main reason to own SMH.
🔗 Further Dayhook ETF research
Compare with AIQ ETF and the AI value chain, SOXX vs SOXL: semiconductor exposure and leveraged risk, and AI chip cycle and portfolio context. Each article addresses a different investment question.
A high-conviction AI chip-cycle satellite holding—not an all-weather diversified ETF. SMH gives a more direct semiconductor supply-chain exposure than thematic AI software funds or broad technology ETFs, with an especially meaningful company-concentration trade-off.
Investment conviction should be proportional to evidence and risk capacity, never to the strength of a headline.
Research notes: Research assembled Nov 16, 2026, U.S. market hours; different holdings, fee and performance data have specifically labeled earlier as-of dates. Primary issuer: VanEck fund documents; dated holdings: holdings and benchmark snapshot; performance: dated fund performance. Commentary and bull/base/bear scenarios are Dayhook Investing analysis, not independently produced analyst targets. Quotes and embeds update after publication. Leveraged exposure, where applicable, is intended for daily results and involves special risks. This is general financial information, not personalized investment, tax or legal advice. ETFs can lose value, including a substantial part or all invested capital. For India-based readers, U.S. dollar and INR foreign-exchange movement, broker access and applicable cross-border rules can affect returns and availability.
