Fund Mechanics

Nvidia Jumped 8.7%. SMH Got 1.9 Points of It. SOXX Got 0.8.

SMH holds Nvidia at 21.70% of the fund. SOXX holds it at 9.05%. Same sector label, same 0.35% expense ratio, and one fund carries more than twice the Nvidia of the other. The weight is the whole story, and it works exactly as hard in reverse.

Nvidia reported results for the quarter ended July 26, and on August 27 the stock closed at $227.98 against a previous close of $209.66, a gain of 8.74%. SMH closed up 3.10% the same day. SOXX closed up 1.95%. Two funds with the same sector label and the same fee, and one captured well over half again as much of the day as the other.

SMH and SOXX carry the same sector label and the same 0.35% expense ratio. SMH held Nvidia at 21.70% of the fund. SOXX held it at 9.05%. Nothing else about the two funds separates them as cleanly as that one number.

The short version

A holding moves a fund by its weight multiplied by its move, and nothing else. Nvidia at 21.70% and up 8.74% is worth about 1.90 percentage points to SMH, which is 61% of the fund's 3.10% day. The same stock at 9.05% inside SOXX is worth about 0.79 points, or 41% of that fund's 1.95%. One company decided most of one fund's session and under half of the other's.

Same Label, Same Fee, Two Different Nvidia Bets

Here is what the two funds actually held, taken from each issuer's most recently published holdings file. Expense ratios are the values ETF BFF records in its own fund data, the same source that renders the SMH and SOXX factsheets.

  SMH SOXX
FundVanEck SemiconductoriShares Semiconductor
Expense ratio0.35%0.35%
Holdings26About 30
Nvidia weight21.70%9.05%
Holdings as ofJuly 30, 2026August 20, 2026
Points from Nvidia's 8.74% move1.900.79
Fund's actual close, Aug 27+3.10%+1.95%
Share of the day from Nvidia61%41%
Points from everything else1.201.16

The two weights are published as of different dates, three weeks apart, and both drift daily with share prices. Treat them as close approximations rather than exact figures for any single trading session. The gap between them is far too large to be an artifact of that timing difference.

The Arithmetic, So It Can Be Checked

A single holding contributes its portfolio weight multiplied by its price move. That is the entire calculation:

Neither fund moved only that amount on the day, because the other holdings moved too, and on a day when one chipmaker reports strong demand the rest of the sector rarely sits still. The point is not the total. It is the share of the total that traces back to one company. For SMH that share was more than twice what it was for SOXX, and the ratio of the two weights, 21.70 divided by 9.05, comes out at about 2.4 times.

This is the same mechanism that governs why one biotech stock doubling barely moves XBI, running in the opposite direction. Equal weighting mutes single names. Concentrated cap weighting amplifies them.

The Same Weight Works Exactly As Hard In Reverse

Concentration is not a feature that only pays out. The 21.70% position that handed SMH about 1.90 points on an 8.74% up day subtracts about 1.90 points on an 8.74% down day. A stock capable of moving close to 9% on a strong report is equally capable of moving close to 9% on a weak one, and the fund's arithmetic does not care which direction it is.

That is the trade being made, and it is worth stating plainly because fund marketing rarely does. A 21.70% weight in the largest company in the sector is a decision to let that company drive roughly a fifth of the fund's day-to-day result. Some people want exactly that. Others want the sector without that much of one name in it, which is closer to what SOXX delivers. Both are defensible positions and neither is the safe default.

Where SOXX Puts the Weight Instead

The twelve points of Nvidia that SOXX does not hold are not left idle. They are redistributed, and where they land describes the fund better than any label does.

SOXX carried Micron at 8.84%, Advanced Micro Devices at 8.02%, Marvell at 5.40%, Intel at 5.12%, Lam Research at 4.30% and KLA at 4.26%. That is a noticeably heavier tilt toward memory manufacturers and the companies that build the equipment chips are made on. SMH's largest positions after Nvidia leaned the other way, toward designers and the foundry that manufactures for them: Taiwan Semiconductor at 9.51%, Broadcom at 6.73%, AMD at 5.43%, ASML at 5.12% and Texas Instruments at 4.95%.

That difference matters on days when the sector does not move as one block. Memory pricing, foundry capacity and lithography equipment orders run on their own cycles, and a fund weighted toward one of them will not track a fund weighted toward another. If memory is the part of the chip cycle you are actually interested in, there is a fund built only for that, and it behaves differently again.

What to Check Before Assuming Any Chip Fund Tracks Any Chip Stock

The general version of this, useful well beyond these two funds:

For the fuller side-by-side on these two, including how their indexes differ in construction, see SOXX vs SMH. For how sector funds concentrate risk in general, the semiconductor ETF guide and the sector ETF guide cover the structural version of this without the news attached.

Frequently Asked Questions

How much Nvidia is in SMH?

Nvidia was 21.70% of the VanEck Semiconductor ETF as of the fund's July 30, 2026 published holdings, making it the largest position by a wide margin. The next largest was Taiwan Semiconductor at 9.51%, followed by Broadcom at 6.73%. SMH holds 26 positions in total and weights them by market capitalisation within a capped methodology, so the weight moves with Nvidia's share price and is re-set at each index review.

How much Nvidia is in SOXX?

Nvidia was 9.05% of the iShares Semiconductor ETF as of its August 20, 2026 published holdings. That is less than half of SMH's weight in the same company. SOXX held roughly 30 positions and spread its top weights more evenly, with Micron at 8.84% and Advanced Micro Devices at 8.02% sitting close behind Nvidia rather than far below it.

Why did my semiconductor ETF not match Nvidia's move?

Because a fund holds many companies and each one contributes only its own weight multiplied by its own move. Nvidia closing up 8.74% contributed roughly 1.90 percentage points to SMH, calculated as 0.2170 multiplied by 8.74%, and roughly 0.79 points to SOXX, calculated as 0.0905 multiplied by 8.74%. SMH closed up 3.10% and SOXX up 1.95% that day, so the remaining 1.20 and 1.16 points came from each fund's other holdings. Those remainders are close to each other because the two funds own many of the same non-Nvidia names.

Does SMH or SOXX cost more to own?

Neither. ETF BFF records both SMH and SOXX at a 0.35% expense ratio, which works out to $35 a year on a $10,000 position in either fund. Fee is not the variable that separates these two funds. The construction of the portfolio is, and the Nvidia weight is the single largest expression of that difference.

Does a higher Nvidia weight mean better returns?

Not automatically. A heavier weight means more of the fund's outcome is decided by one company, and that holds in both directions. The same 21.70% position that added roughly 1.90 points to SMH on an 8.74% up day would subtract roughly 1.90 points on an 8.74% down day. Which construction produced the better result over any particular window depends entirely on what Nvidia did relative to the equipment makers and memory manufacturers that carry more weight inside SOXX, and that relationship has reversed more than once.

What else is different between SMH and SOXX besides Nvidia?

The weight SOXX does not put into Nvidia is redistributed rather than left idle, and where it goes describes the fund. SOXX carried Micron at 8.84%, Advanced Micro Devices at 8.02%, Marvell at 5.40%, Intel at 5.12%, Lam Research at 4.30% and KLA at 4.26%, which is a heavier tilt toward memory manufacturers and equipment makers. SMH's top positions after Nvidia leaned toward chip designers and the foundry that builds for them, with Taiwan Semiconductor at 9.51%, Broadcom at 6.73% and ASML at 5.12%.

The Expense Report Newsletter

One ETF concept a week. Free, forever.

Plain-English ETF breakdowns like this one, straight to your inbox. No jargon, no stock tips.

Reviewed by a CFA® charterholder · No spam · Unsubscribe anytime

How ETF fees, structure and holdings actually work, without the sales copy.

Start with ETF basics →
Portfolio weights and holding counts quoted here come from each issuer's published holdings file on the dates stated in the table, are point-in-time figures, and drift daily with share prices between index reviews. The two funds publish on different schedules, so the weights compared here are dated three weeks apart. Published holding counts for SOXX vary between roughly 30 and 34 depending on the data source and whether cash and derivative line items are included, so the count is given here as an approximation and the Nvidia weight, which is consistent across sources, carries the argument. Expense ratios are the values ETF BFF records in its own fund data, the same source that renders the factsheets linked above, and issuers change them. Verify all figures against the fund's own materials before acting on any of them. The 1.90 and 0.79 point figures are arithmetic on those stated weights, shown so they can be checked, and describe the contribution of one holding to one day's fund return. Closing prices and percentage moves for August 27, 2026 are historical single-session figures and are used here only to make the weight arithmetic concrete. Because the published weights predate that session, the contribution figures are close approximations rather than exact attributions. Sector and industry funds concentrate risk in a way broad index funds do not, and a fund with more than a fifth of its assets in a single company carries the outcome of that company in both directions. Past performance does not guarantee future results. ETF BFF is not a registered investment adviser, and nothing here is personalized financial advice or a recommendation to buy, sell or avoid any security.