SOXX vs SMH vs XSD: Semiconductor ETF Holdings, Overlap, and Basket Analysis

The short answer is basket construction. SOXX follows the NYSE Semiconductor Index, which selects 30 U.S.-listed securities and applies float-adjusted market-cap weights with caps. SMH follows the MVIS US Listed Semiconductor 25 Index, which emphasizes large and liquid U.S.-listed semiconductor businesses and allows larger weights near the top. XSD follows the S&P Semiconductor Select Industry Index, which starts from a broader U.S. total-market semiconductor universe and reweights constituents toward equal weights, subject to liquidity limits.

That distinction appears in the dated top-ten totals: SOXX was between the other two, SMH was the most concentrated in its ten largest positions, and XSD spread more of its weight below the top ten. “More holdings” or “lower top-ten concentration” does not remove the common industry exposure. The three funds can still respond to the same semiconductor demand, capital-spending cycle, inventory changes, and company news.

This comparison answers how the baskets are built and where their reported names overlap. It does not rank recent returns or recommend a fund. The holdings below are aligned to September 24, 2026 so the comparisons use one date. Check the issuers’ current holdings pages before relying on any ticker weight because prices, index membership, and fund positions change.

SOXX seeks to track the NYSE Semiconductor Index. Its index document describes 30 securities ranked by security-level float-adjusted market capitalization, selected from eligible U.S. listings. The fund’s page reports the NYSE Semiconductor Index as its benchmark. The index is therefore a capped large-company basket; a semiconductor company can fall outside it because it is not among the selected securities or does not meet the index’s eligibility rules. See the iShares SOXX profile, the iShares index description and SAI, and the September 2026 SOXX index supplement.

SMH tracks the MVIS US Listed Semiconductor 25 Index. MarketVector describes it as the 25 largest and most liquid U.S.-exchange-listed semiconductor companies. New constituents generally need at least 50% of revenue from semiconductor production or equipment; current constituents can remain with a lower 25% threshold. That can admit U.S.-listed foreign issuers such as Taiwan Semiconductor Manufacturing, ASML, and SK Hynix securities. A U.S. listing requirement is not the same as a U.S. corporate-domicile requirement. See the VanEck SMH profile and the MarketVector index guide.

XSD tracks the S&P Semiconductor Select Industry Index. The index draws from the semiconductor sub-industry within the S&P Total Market Index, which spans a broader U.S. equity universe than the large-cap slice alone. Its membership screens include investability and liquidity requirements. This is why XSD can include smaller semiconductor names that do not appear in SOXX or SMH. The State Street XSD page describes the benchmark and the S&P Select Industry methodology documents its screens and weight process.

The currently posted iShares SAI, dated July 31, 2026 and supplemented on September 2, states the general NYSE Semiconductor Index weighting limits: 8% for each constituent, 4% for positions outside the initial five largest, and 10% in aggregate for ADRs when index weights are set. These are scheduled index-weight limits, not continuous ceilings on every daily fund holding. The September 24 fund file shows Intel at 10.20%, AMD at 9.58%, and Micron at 8.08%. That observed difference deserves attention, but the holdings file alone does not show the index’s reset weights, the fund’s implementation timing, or the cause of the daily weights; it is not enough to conclude that the index rule was breached. See the iShares SAI and its September 2 supplement.

The supplement also sets annual reconstitution after the third Friday in September using end-August inputs, and adds Fast Entry for qualifying new listings that rank in the top 15. A temporary 4% individual cap and 10% aggregate ADR limit apply to a qualifying Fast Entry addition only until the next reconstitution or rebalance, with timing exceptions. These additional Fast Entry limits do not replace the general scheduled weighting rules and do not explain the observed September 24 weights by themselves.

SMH also uses a modified float-adjusted market-cap approach, but its rulebook permits a “Large-Weights” group with individual weights from 5% up to 20%, while the remaining “Small-Weights” group has a 4.5% maximum. The split makes room for several large companies to occupy meaningful positions at once. It does not guarantee that any one stock will sit at the maximum: the actual weight depends on index eligibility, market capitalization, free float, and the redistribution process. The large-weight group is capped at 50% in aggregate. Companies with less than 50% exposure to qualifying semiconductor activities are capped at 20% in aggregate. These are scheduled index weighting rules; daily fund weights can drift.

XSD starts from equal weights at each quarterly index rebalancing, then adjusts them for liquidity and float-adjusted market capitalization. S&P’s methodology has a special case for an index with exactly 22 constituents that uses equal weights without the basket-liquidity constraints; that case does not apply to this September 24 snapshot, which had 47 index constituents. Under S&P’s updated methodology, a constituent’s base maximum is the lowest of 4.5%, three times its three-month median daily value traded divided by the theoretical portfolio value (TPV), and 4.5% of its float-adjusted market capitalization divided by TPV. If all constituents are capped and their total weight still falls short of 100%, S&P relaxes the liquidity multiplier in 0.1 increments, then the single-stock cap in 0.1-percentage-point increments, and then TPV downward in $0.1 billion increments until a feasible result is found. The single-stock cap’s upper limit in that process is 4.8%. A separate check on the third-to-last business day of March, June, September, or December resets index weights if constituents above 4.8% together exceed 50%, using that quarter’s previously calculated weights. These are index rules, not a promise that every XSD fund holding will remain equal or below the reset weights: share-price changes, cash, and fund implementation can move daily holdings between resets. See the S&P Select Industry methodology and S&P’s 2024 implementation announcement.

These are not three versions of the same weighting formula. SOXX limits a market-cap-weighted basket; SMH can assign much more weight to its largest eligible companies; XSD gives smaller constituents more influence through its modified equal-weight reset. A fund’s name alone does not tell you whether it is concentrated in mega-cap leaders or spreads more weight across the industry.

The following issuer snapshots use the same date. Each cell shows company, ticker, and percentage of net assets. Top-ten sums add the displayed weights, rounded to two decimal places.

SOXX’s holdings file is published on the iShares profile and CSV download; SMH’s dated table is on VanEck’s holdings page; and XSD’s daily positions are available from XSD daily holdings workbook. For XSD, the issuer reported 48 fund holdings and 47 index constituents on the date; cash and implementation lines are not additional operating companies. For SMH, use the issuer’s total-holdings field rather than assuming every line is a company stock.

On the displayed September 24 weights, SMH’s top ten accounted for 67.62% of net assets, SOXX’s for 61.73%, and XSD’s for 27.14%. The largest single weight also differed: NVDA was 19.28% in SMH, INTC 10.20% in SOXX, and AMD 2.80% in XSD. These measures explain why the three funds can have different company-level sensitivity even while targeting the same industry.

Top-ten concentration is a narrow statistic. It says nothing by itself about volatility, losses, valuations, or whether a lower-concentration basket is safer. XSD still holds a sector-specific equity portfolio; a broad spread across semiconductor companies can remain exposed to a downturn in the same industry. Conversely, a larger top-ten share does not prove a fund will move more than another fund in every period.

The reported number of holdings also needs context. A fund can have dozens of company lines, but many may have small weights. Compare the weight distribution and the business roles of the positions rather than treating “48 holdings” as if every name contributed equally. The index constituent count, fund holding count, cash lines, and total number of rows in a downloadable file are related but not interchangeable measures.

For the September 24 snapshot, common securities were matched by their exact reported ticker symbols. The right-hand columns add the weights of those shared ticker symbols within each fund. Common shares and depositary receipts are included; cash, derivatives, and money-market positions are excluded. In QQQ’s file, ASML appears under ticker ASML and security type DRNY; we treat this equity depositary receipt as an equity holding and include it in the exact-ticker match. We add each source’s available precision before rounding the reported totals. This measures shared disclosed basket weight; it is not a correlation estimate or a forecast of how prices will move together.

This table adds context to the top-ten comparison. SOXX and SMH shared most of each fund’s weight in names with the same ticker, even though SMH was more heavily concentrated in NVDA. XSD shared fewer of its own portfolio dollars with the other two, in part because its equal-weight design assigned more weight to smaller companies. A different listing symbol can also conceal an economic relationship: SOXX lists SK Hynix as SKHY, while SMH reports SKHYV. The exact-ticker method counts those as different securities. It does not attempt to merge ADRs, local listings, or different share classes into one issuer-level exposure.

A separate symmetric measure takes the smaller weight for each matched ticker and adds those minima: sum(min(weight_A, weight_B)). On the same source files it equals 66.70% for SOXX–SMH, 37.02% for SOXX–XSD, and 27.07% for SMH–XSD after rounding. This answers how much portfolio weight can be paired at matching ticker weights, whereas the table above answers how much each fund allocates to the shared set. The two directional percentages need not be equal. Neither method joins SKHY and SKHYV or estimates return correlation.

QQQ follows the Nasdaq-100, a multi-industry group of large non-financial companies listed on Nasdaq. XLK tracks information-technology companies drawn from the S&P 500. Semiconductor ETFs are narrower industry baskets, but their largest U.S.-listed constituents can already be present in both broader funds. The following exact-symbol matches use September 24 holdings from the issuers.

XLK’s own September 24 sector breakdown reported 45.83% in semiconductors and semiconductor equipment. Its basket also contains software, hardware, and other information-technology industries, so XLK is not a pure semiconductor fund. QQQ is broader still: it includes several industries outside information technology. The holdings overlaps above show that these wider funds can already carry material semiconductor-company weight without being dedicated semiconductor funds. Sources: Invesco QQQ information, QQQ holdings data, Nasdaq-100 methodology, and XLK daily holdings workbook.

The holdings span different activities. NVIDIA and AMD design processors; TSMC fabricates chips for customers; Micron and SK Hynix make memory; ASML, Applied Materials, Lam Research, and KLA supply tools used in chip production and inspection; and companies such as Cadence and Synopsys sell electronic-design software. This is a simplified map: companies can operate across more than one area, and each ETF’s index classifies eligible securities according to its own rules.

That distinction matters because semiconductor revenue does not respond to one identical driver. Chip designers can depend on product demand and customer adoption. Foundries and memory producers invest in capacity and can face changing utilization or supply-demand cycles. Equipment suppliers depend on customers’ capital spending and installation plans. An index can hold all of these businesses, but the exact mix differs by eligibility rules and weights. The holdings table cannot substitute for company-level financial analysis.

SMH’s global-listed holdings can add foreign-domiciled issuers through U.S. listings, while SOXX and XSD are also U.S.-listed equity baskets with their own eligible securities. A U.S.-listed ADR still carries business, currency, custody, governance, and geopolitical exposures tied to its issuer and markets. Listing venue alone does not make an international semiconductor business economically domestic.

Current issuer materials list annual expense ratios of 0.33% for SOXX, 0.35% for SMH, and 0.35% for XSD. On an unchanged $10,000 balance for one year, the simple arithmetic is:

• SOXX: $10,000 × 0.0033 = $33
• SMH: $10,000 × 0.0035 = $35
• XSD: $10,000 × 0.0035 = $35

In this static example, SMH and XSD cost $2 more than SOXX per $10,000 per year. It is not a bill or a return comparison: fund expenses accrue against changing assets, and the calculation excludes brokerage charges, bid-ask spreads, taxes, market impact, and tracking differences. A lower fee does not imply a better-fitting basket. Recheck the current SOXX fee schedule, SMH expense ratio, and XSD prospectus information because fees and terms can change.

If an account holds more than one ETF, add the underlying security weights using the amounts invested. For a company held at weight (w_A) in fund A and (w_B) in fund B, with dollar allocations (a) and (b), its combined account weight is (a × w_A + b × w_B) ÷ (a + b). This is a portfolio arithmetic step, not a statistical measure of diversification.

The shared-ticker table can also be used for a simple example. Suppose an account allocates $5,000 to SMH and $5,000 to QQQ, using the September 24 snapshot and the rounded table percentages. The names that appeared in both funds represented 84.83% of SMH and 33.83% of QQQ, so the amount assigned to the shared ticker set would be $5,000 × 0.8483 + $5,000 × 0.3383 = $5,933, or 59.33% of the $10,000 ETF allocation. This assumes the same holdings snapshot and unchanged allocations; it is not a current portfolio report.

Do not interpret that $5,933 as a prediction that 59.33% of the account will rise or fall together. It simply totals the portfolio dollars allocated to ticker symbols present in both dated funds. Company prices can move differently, fund weights change, and exact-ticker matching treats distinct listed securities separately. To know the current amount, use current weights and actual account positions; then add exposures across any other funds or stocks in the account.

All three ETFs concentrate on the semiconductor industry, whose companies face technology shifts, product cycles, large fixed investment, customer concentration, trade restrictions, and swings in supply and demand. A company can gain from strong chip demand and still face pricing pressure, execution problems, or a change in customer needs. Index rules do not prevent those business risks.

The September 24 values are a dated holdings comparison, not live portfolio weights. The fund and index pages may update on different schedules, and the issuers’ portfolio accounting can include cash, derivatives, or other non-stock lines. The analysis does not calculate historical correlation, volatility, valuation, maximum drawdown, tax consequences, or expected return. It also does not decide which ETF belongs in an individual account.

For a related broad-technology comparison, see XLK vs. VGT vs. FTEC. For Nasdaq-100 overlap with the S&P 500, see QQQ vs. VOO. The useful sequence is to read the index rule, inspect holdings measured on one date, calculate the share of account dollars assigned to common companies, and then consider how those companies fit the rest of the portfolio.