Nvidia is 6.6% of every ETF that holds it — and missing from three quarters of them

Ivan — Wed, 16 Sep 2026 21:57:54 +0000

Most "what do ETFs own" numbers are averages across every fund, which drags each company toward zero: three quarters of US equity ETFs hold no Nvidia at all, so Nvidia's average across all funds is a meaningless 1.6%.

The useful question is what a fund holds when it holds the company at all. On that measure, as of 15 September 2026 across 964 funds — every US-listed equity ETF we track that is unlevered, at least three years old and holds at least $20M:

#	Company	Average weight when held	Funds holding it	Funds holding it above 10%
1	Nvidia	6.6%	237	43
2	TSMC	6.6%	106	28
3	Apple	5.8%	220	15
4	Alphabet	5.0%	235	21
5	Microsoft	5.0%	229	13
6	Samsung Electronics	4.9%	76	5
7	SK Hynix	3.9%	79	3
8	Amazon	3.8%	208	6
9	Tencent	3.4%	69	5
10	Meta	2.8%	211	5

Companies held by fewer than 25 funds are excluded, so a single concentrated fund cannot put a company at the top.

Why this list is not the mega-cap list

Five of the ten make semiconductors, the memory beside them, or the phones they go into: Nvidia, TSMC, Samsung Electronics, SK Hynix and Apple. Tesla, larger than most of this table, does not appear at all — it averages 2.2% where it is held.

The reason is that chip companies are held by funds built to concentrate — semiconductor and technology funds with 25 to 40 holdings — while a company like Meta is mostly held by broad funds that spread across hundreds of names, which is why it ranks tenth despite sitting in 211 of them. TSMC makes the point twice over: it is in only 106 funds, a ninth of the market, yet ranks second, because the funds holding it hold it hard.

How far does concentration go?

The averages hide the tail, and the tail is where the risk sits. Counting funds by how much of their weight goes to one company:

Company	Held by	Above 5%	Above 10%	Above 20%
Nvidia	237 funds	147	43	2
TSMC	106 funds	50	28	2
Apple	220 funds	136	15	0

So 43 funds put more than a tenth of everything they own into Nvidia, and 28 do the same with TSMC — a company more than a quarter of whose holders are that concentrated in it.

And at the fund level?

Our own exposure pages, 15 September 2026.

Magnificent 7 exposure — average fund 7.4%, VOO 33.2%:

Fund	Exposure
MGK — Vanguard Mega Cap Growth	59.5%
VUG — Vanguard Growth	56.5%
XLG — Invesco S&P 500 Top 50	54.1%
IYW — iShares U.S. Technology	52.7%
SPYG — SPDR Portfolio S&P 500 Growth	51.6%

Technology-sector exposure — average fund 23.6%, VOO 45.2%:

Fund	Exposure
PSI — Invesco Semiconductors	100.0%
SMH — VanEck Semiconductor	100.0%
SOXQ — Invesco PHLX Semiconductor	100.0%
SOXX — iShares Semiconductor	99.9%
FTXL — First Trust Nasdaq Semiconductor	99.7%

Three funds are entirely technology to one decimal place. A fund at the top of the Magnificent 7 table puts more than half its weight in seven companies, against 33.2% for an S&P 500 tracker — which is itself a third in those seven.

How it was computed

The universe is every US-listed equity ETF with at least three years of history, at least $20M in assets and no leverage — no leveraged or inverse products, no bond, commodity or currency funds, no cash or money market funds. That comes to 964 funds on 15 September 2026.

Each listing is identified by its ISIN, then dual listings and share classes are merged into one company — Alphabet's A and C shares, TSMC's Taiwan listing and its ADR — so a company is counted once per fund.

"Average weight when held" is the mean across only the funds that hold the company. The all-fund average, for comparison, is 1.6% for Nvidia and 0.7% for TSMC.

Where these numbers stay current

ETF Copilot recomputes this every day: etf-copilot.com/exposure.

Does SCHD overlap with VTI? I measured every holding

Ivan — Fri, 28 Aug 2026 18:37:05 +0000

On a look-through of all 98 holdings as of 9 September 2026, 100% of SCHD's weight sits inside VTI. Next to a total-market fund, Schwab U.S. Dividend Equity ETF adds almost no new companies — it re-weights ones you already own.

Every figure here is dated 9 September 2026. ETF Copilot recomputes them daily, so the current numbers are on the live SCHD look-through.

I worked through SCHD's holdings and compared them against a thousand-plus US-listed equity ETFs. Three findings, none of which appears on any published fund profile I could find.

SCHD, as of 9 September 2026


Share of SCHD's weight already inside VTI	100% — 96 of its 98 companies are in both
Share of VTI inside SCHD	7%
Trailing P/E of its holdings	19.0× — the cheapest of its four largest peers
Against its own record since 2017	More expensive than 82% of its monthly readings
Companies carrying half the fund	13, of 98 holdings

How much does SCHD overlap with VTI?

100% of SCHD's weight sits inside VTI as of 9 September 2026 — 96 of its 98 companies are in both, and the two that are not are too small to move the rounding. Against VT it is 99%, against VYM 99%, and against SCHX 97%. Going the other way, only 7% of VTI is inside SCHD — it sits inside the broad fund rather than beside it.

SCHD's weight already held by	Share
VTI	100%
VT	99%
VYM	99%
SCHX	97%
VIG	55%

That last row is why the measure is worth trusting. VIG is the same category and the same dividend framing, and it holds only 55% of SCHD — VIG and SCHD are genuinely different portfolios; SCHD and VTI are largely the same one, re-weighted. The calculation discriminates rather than simply debunking everything it touches.

What does SCHD add to a portfolio that already holds VTI?

A tilt in weighting, not additional diversification: as of 9 September 2026, 100% of SCHD's weight is inside VTI, so SCHD adds essentially no companies that are not already in VTI. Whether a dividend-quality tilt is worth having is a separate question, and a reasonable one. It is just not a diversification question.

Is SCHD expensive right now?

Both answers are true at once, and this is the finding I care about most.

Against its peers, SCHD is cheap. Rolling every holding's trailing earnings up to fund level, SCHD traded at 19.0× trailing earnings on 9 September 2026 — the cheapest of its four largest peers, which run up to 25.3×.

Fund	Trailing P/E, 9 Sep 2026	Its own median	More expensive than
SCHD	19.0×	17.1×	82%
VYM	20.7×	17.4×	91%
DGRO	22.6×	19.0×	92%
CGDV	25.2×	25.0×	52%
VIG	25.3×	23.4×	80%

"More expensive than" is the share of that fund's own monthly readings sitting below today's.

Against itself, SCHD is expensive. As of 9 September 2026, SCHD's trailing P/E of 19.0× is more expensive than 82% of its own 114 monthly readings since 2017 — its median over that period is 17.1×.

Funds screening for dividend growth select companies the market already prices richly, so the whole peer group re-rates together: four of the five are more expensive than at least 80% of their own history. The comparison table moves as one and tells you nothing about level.

Only one of those two numbers appears on a typical comparison table. The peer view and the historical view point in opposite directions, and the view almost everyone uses is the one that cannot see it.

Is SCHD concentrated?

Yes — 13 of SCHD's 98 companies carry half the fund, and its ten largest are 42% of it, as of 9 September 2026. On effective holdings — the equally-weighted count that would produce the same concentration — SCHD behaves like a 34-stock portfolio, more concentrated than 86% of US dividend funds.

"98 holdings" is true and tells you almost nothing.

Why none of this is on a fact sheet

Not because anyone is hiding it. Each of these figures requires looking through the fund to the companies underneath, mapping those companies to a stable identity so they can be matched across funds, and recomputing as holdings change. A fact sheet is a snapshot of one fund; these are relationships between funds, and they move.

On ETF Copilot's nightly recomputation across the US-listed equity universe — which is where the peer-group and percentile figures above come from — doing this once for one fund is an afternoon's work. Doing it every night for all of them is the part that took a while.

Check it on any pair you own

Take two funds you own together and ask what share of the smaller one's weight already sits inside the larger. Above 95% and you own one position with extra steps. At 55%, like SCHD against VIG, you own two things.

I build ETF Copilot (https://etf-copilot.com), disclosing that it's mine. Every figure above is from its nightly recomputation, dated 9 September 2026, and is checkable without an account on the live SCHD look-through, which always carries the current numbers.

DEV Community: Ivan