You think you bought an index fund. What you actually bought is a concentrated bet on roughly eight companies wearing an index fund's name tag.
The wrapper is honest. QQQM is an exchange-traded fund, a basket of stocks that trades like a single share. It holds every company in the Nasdaq-100 Index in roughly index proportions.
Eight of those positions carry about 41% of the money. Twelve carry 52%, and the top fifty carry nearly 88%. The rest is rounding, and the rounding is not where your return comes from.
None of this is a scandal or a hidden fee. The Nasdaq-100 follows a published rulebook, and this fund replicates that rulebook cheaply. The surprise lives in what the rules produce, because the rules were rewritten in May 2026 to admit winners faster.
Four of its ten largest positions are semiconductor or AI-compute names, and the memory cycle has rewritten the top of the book in twelve months. Micron, a 480% gainer over that year, now ranks fourth in the fund.
What follows is the ownership map, the machinery, the bill, the scoreboard, and the argument a short-seller would make. If you hold this fund in a retirement account and have never read its index methodology, you are in very large company.
What you actually own
Start at the top, with weights as of 14 September 2026. NVDA is 8.23% of the fund, AAPL 7.88%, MSFT 6.05%, MU 4.64% and AMZN 4.39%.
AMD takes 3.58%, GOOGL 3.28%, META 3.18%, GOOG 3.04% and TSLA 2.89%. Those ten add up to 47.15%. Alphabet appears twice because the company has two share classes, which is one reason the fund reports 106 holdings for fewer companies than that.
The concentration math steepens down the list. Eight names account for 41.2% of assets, twelve for 52.2%, the top fifteen for 58.19% and the top fifty for 87.86%. The remaining fifty-six positions split about 12%.
Sector exposure, as of 31 August 2026, is what you would guess. Technology is 66.36% of the fund on Invesco's classification. Under the GICS scheme the technology number falls near 54%, because Alphabet and Meta are filed as communication services.
The index excludes financial companies and real estate entirely, so both sit at 0.00% in every sector table. You own no bank and no landlord here, which sounds clean until you notice what it removes from a broad portfolio.
Nor is this a purely American list. Shopify and Thomson Reuters are Canadian, PDD is domiciled in the Cayman Islands, and Linde and Seagate are Irish. ASML, Ferrovial, Nebius and NXP are Dutch, while Arm Holdings and Coca-Cola Europacific Partners are British.
Non-technology weight is real but modest. On Invesco's June 2026 sector table, consumer discretionary runs about 16.4%, mostly AMZN and TSLA. Staples contribute a little through WMT and COST, health care sits near 3.6% and industrials near 3.1%.
The past year sorted this book violently. From 17 September 2025 to 16 September 2026, on total return, MU gained about 480%, INTC 306%, MRVL 224% and AMD 222%. AAPL added 40%, GOOGL 38% and NVDA 26%.
The laggards sit in the same wrapper. Over that same year APP fell 46%, PDD 42%, NFLX 38%, ADBE 31%, BKNG 22%, TSLA 16%, META 13%, MSFT 3% and AVGO 1%.
How the sausage is made
Nobody picks these stocks. QQQM is passive, which means Invesco's job is to own what the Nasdaq-100 owns, in the same proportions, as cheaply as possible. The fund keeps at least 90% of total assets in index securities, and it holds all 106 names rather than a statistical sample.
The index is modified cap-weighted. Cap-weighting means bigger companies get bigger slices, so a fund's return is decided by its largest holdings. The word modified means the raw maths gets bent by a rulebook.
That rulebook weights each company by its full listed market capitalisation, not the float-adjusted figure most index providers use. Every listed share counts, including shares insiders hold. Eligibility is narrow: a Nasdaq listing on the Global Select or Global Market tier, and no financial companies.
Then come the caps. At company level, any weight above 24% forces a reset so nothing tops 20%. A cohort of companies above 4.5% is cut to 40% if it would reach 48% together.
At security level, a position above 15% is capped at 14%. The five largest are trimmed to 38.5% if they sum to 40%, and anything outside the top five is limited to the lesser of 4.4% or the fifth-largest weight. The maths repeats until every test passes.
Rebalancing runs quarterly, in March, June and September. The reference date is the last trading day of the prior month, and the changes take effect at the open of the first trading day after the third Friday.
A breach of the 24% or 48% tests triggers a special rebalance in between. Every December the index is reconstituted, which is the trade word for rebuilding the membership list. The announcement lands after the close on the sixth trading day before the effective date.
May 2026 changed the machinery. A three-times free-float cap replaced the old rule requiring at least 10% of shares in public hands. A Fast Entry rule now admits a large new listing quickly, still subject to that float cap.
The old 10 basis point test and intra-quarter share-count updates are gone. Names ranked outside the top 125 by full market value are removed at the quarterly reviews.
The wrapper does real work too. ETFs create and redeem shares in kind, swapping baskets of stock rather than cash, so the fund rarely sells securities to meet redemptions. That suppresses capital gains distributions, which is the tax efficiency pitch.
Shares trade all day at market prices, so you get intraday liquidity. The fund family's prospectus also permits securities lending, the practice of lending positions to borrowers for a fee, up to a third of total assets and continuously collateralised.
The bill
The expense ratio is the annual fee the fund skims from your money, and QQQM charges 0.15%. On a $10,000 position that is about $15 a year, or roughly $150 over a decade before compounding. One published model puts the compounded ten-year cost near $298.
The obvious comparison is QQQ, the sister fund on the identical index. Its fee fell from 0.20% to 0.18% on 22 December 2025, when QQQ converted from a unit investment trust into an open-end fund. The gap is now 3 basis points, roughly $3 a year per $10,000, or about $59 across a decade in that same model.
The uncomfortable comparison is newer. SPDR's QNDX launched on 23 June 2026 at 0.10%. iShares' IQQ followed on 9 July 2026 at 0.12% gross and 0.10% net under a waiver running through 31 July 2027. Both are small, both sample the index instead of owning all of it, and both trade around 0.04% spreads.
Fund economics reach you in smaller ways too. QQQM's median bid-ask spread was 0.01% as of 4 September 2026, against 0.00% for QQQ. Its closing premium or discount to net asset value was flat in late August.
A premium means the shares cost more than the basket underneath; a discount means less. One intraday snapshot caught a 0.19% quoted spread, which is the price of impatience in a fund trading about 2.2m to 2.6m shares a day against QQQ's 30.7m.
Tracking difference, the gap between fund and index after all costs, is small enough that fees decide the argument.
The scoreboard
Judge a fund on what it actually paid you, after costs, and label every window. To 16 September 2026, QQQM's total return was 15.03% so far in 2026 and 19.85% over one year. Over three years it compounded at 24.66% a year, and over five at 14.02%.
Since inception in October 2020 it has compounded at 16.60% a year, for a cumulative 148.5%. QQQ over the same windows returned 14.99%, 19.79%, 24.55% and 13.93%, and the index itself 14.63%, 19.26%, 23.91% and 13.28%.
Invesco's own standardized NAV figures, as of 31 July 2026, flip the ordering of fund and index. QQQM returned 12.28% year to date, 22.38% over one year, 22.23% a year over three, 14.30% over five and 16.52% since inception.
The index on the same basis: 12.38%, 22.58%, 22.42%, 14.46% and 16.68%. That gap of roughly 0.1 to 0.2 points a year is mostly the fee.
The broader market lost this race. To 16 September 2026, the S&P 500 returned 11.17% year to date, 15.53% over one year, 20.69% a year over three and 12.53% over five. QQQM beat it on every window.
Valuation is where you pay for that. The Nasdaq-100's trailing price-to-earnings ratio was 35.24 at the end of June 2026, and the S&P 500's was 24.36 in mid-September. A trailing multiple is the price you pay for each dollar of past profit, so the index carried a premium in the mid-thirties to mid-forties percent range.
The forward multiple is kinder. The index traded at 20.56 times expected earnings on 15 September 2026 against a five-year median of 24.60, around the 7th percentile of that range. The S&P 500 sat at 19.11 on 11 September, narrowing the premium to about 7.6%. Estimates for the same index ran from 21.07 to 25.17 across providers.
Earnings are the reason the multiple holds up. Forward twelve-month earnings growth for the index was estimated at 47% on 31 July 2026, up from 29% in December 2025. Index earnings per share rose 45% year over year in the latest quarter, with blended growth near 75% once unrealised Anthropic stakes at Amazon and Google are counted.
Income is an afterthought. QQQM's trailing distribution yield is about 0.45%, and the 30-day SEC yield was 0.46% at the end of June. The S&P 500 yields roughly 0.98%.
The bet, and the bear case
Strip away the structure and one wager remains: the biggest listed technology companies keep converting capital spending into earnings faster than their multiples expand.
The fund is a geared expression of that idea, because winners get heavier every quarter and the heaviest names are paid first by the AI buildout.
No cap has restrained it yet. The largest position in the fund is 8.23%, nowhere near the 24% company limit, so the deconcentration rules sit unused. Weight comes down through price or not at all, and if the leaders fall the fund's largest holdings are exactly what is falling.
The rebuild schedule has a habit of buying strength. Each December, additions replace deletions, and the 2025 vintage had already run hard beforehand: WDC was up 284%, STX 225% and INSM 152%.
The names pushed out fell after they left, with TTD down 68%, LULU 46%, CDW 21%, GFS 19% and ON 14%. Nasdaq's own research on adds and deletes since 2010 found deletions beat additions in the following year, 17.8% against under 12% on average.
The June 2026 review sharpened the tilt. Out went a cable operator, an IT services firm, an insurance data provider, a biotech and a cybersecurity name. In came five compute and space listings: ALAB, CRWV, NBIS, RKLB and TER. That is an index choosing the cycle, not the cycle choosing the index.
SpaceX is the live case study. It entered through Fast Entry at a weight of roughly 1.25%. Analysts expect that to rise toward 1.51% at the 21 September 2026 rebalance, with one pro forma estimate near 2.82%.
The mechanical buying behind it is estimated at $12.4bn to $15.5bn, days before roughly 2.3bn shares leave lockup across October and November. The name is priced near 104 times trailing sales.
Memory drives the middle of the book now. MU's 4.64% weight is a direct output of a cycle that produced a 480% year, and NVDA's 8.23% rests on hyperscaler capital budgets. A digestion phase in that spending does not dint the fund evenly; it lands on the positions that decide the fund's return.
The multiple is the second half of the bear case. The forward multiple sits in the 7th percentile of its five-year range, which a bear reads as the market already paying for the growth. Against the S&P 500, the trailing multiple carries a mid-thirties percent premium.
Asymmetry is not theoretical. In July 2026, QQQM's net asset value fell 6.60% while the S&P 500 fell 0.06%. One month proves nothing, but it previews what a beta near 1.23 feels like when leadership stumbles.
Even the value case thinned. QQQM was the cheap door into the Nasdaq-100 at 0.15% against QQQ's 0.20%. QQQ now charges 0.18%, the gap is 3 basis points, and QNDX and IQQ undercut both at 0.10%.
The prospectus calls the fund non-diversified, which is regulator-speak for permission to concentrate. Its liquidity is a fraction of QQQ's, 2.2m to 2.6m shares a day against 30.7m, which is the wrong kind of difference on a violent day.
What would prove the bears right? The easy tell is the top-ten weight, now 47.15%, breaking above 50%. The harder tell is earnings. If that 47% forward growth estimate is revised below 20% while the forward multiple holds at 20 times or higher, the premium has nothing left to justify it.
How to use it
Treat this as a core holding wearing satellite clothing. If you already own a market-wide fund, QQQM is a second helping of the same seven or eight companies rather than a diversifier. Check the overlap before you add it.
It suits an investor with a long horizon, a high tolerance for drawdown and a deliberate view that listed technology keeps compounding faster than the broad economy. It suits someone who wants Nasdaq-100 exposure in a taxable account and cares about a few basis points of drag.
It does not suit anyone who needs income, because the yield is under half a percent. It does not suit a portfolio already tilted toward growth through an S&P 500 fund, because the marginal diversification is thin.
As a replacement, it can stand in for a Nasdaq-100 allocation or for the growth sleeve of a wider portfolio. As a substitute for genuine diversification, it does not work.
The verdict
QQQM is a well-built, cheap wrapper around a bet that has worked for six years and is now priced for it. Own it as a deliberate concentration, not as the passive default.
Nothing in the structure is broken. What changed is the price of admission and the number of companies doing the work.
What to watch next: the 21 September 2026 rebalance, which prints SpaceX's new weight and the buying it forces. The November release of SpaceX shares from lockup, which tests whether the index bought that winner at the top or in the middle.
The next earnings round for MU and NVDA, where that 47% growth estimate either holds or breaks. And the top-ten weight itself, which tells you whether the concentration is still creeping.
Disclosure: The Signal holds no position in QQQM. Positions may change. This is not financial advice.
