History

THE 10 LARGEST STOCKS NOW MAKE UP AROUND 40% OF THE S&P 500

MARKET HISTORY · 2026-09-28 · index weights as of September 23, 2026
BY RYAN, FRANK & DILLON · MARKET MAULERS FOUNDERS

Put $100 into an S&P 500 index fund today and about $40 of it goes to ten companies. The other $60 gets spread across the remaining 490. Most people who own "the market" have no idea they are this concentrated, and most people who trade ES have never sat down and worked out what it means for the contract they are clicking on.

So we did. This piece covers what the top ten actually is right now, how it got there, what happened the last two times the index got this top-heavy, and the honest argument that this time the weight is earned.

The Actual Ten, As Of September 23

These are the weights inside SPY, the largest S&P 500 ETF, which tracks the index. One detail that trips up almost every "top ten" chart: Alphabet has two share classes in the index, GOOGL and GOOG. Count them separately and you get ten tickers worth 38.7%. Count Alphabet once, as the single company it is, and the ten largest companies come to 40.2%.

#	Company	Weight	Running total
1	Nvidia	8.22%	8.22%
2	Apple	7.41%	15.63%
3	Microsoft	5.60%	21.23%
4	Alphabet (GOOGL + GOOG)	5.38%	26.61%
5	Amazon	3.68%	30.29%
6	Broadcom	2.54%	32.83%
7	Meta	2.47%	35.30%
8	Micron	1.82%	37.12%
9	Tesla	1.61%	38.73%
10	AMD	1.51%	40.24%

We cross-checked against a second, independently published set of index weights and got 40.4%, with Berkshire Hathaway sneaking in at number ten instead of AMD. That difference is worth knowing about. The top five are stable. Spots six through ten reshuffle with a few percent of price movement, so any list you see of "the top ten" is a snapshot of one afternoon.

Look at the first two rows again. Nvidia and Apple together are 15.6% of the index. On the eve of the dot-com collapse in 2000, the two largest names, Microsoft and General Electric, held a combined 9.1%.

Eight of the ten are the same trade. Nvidia, Microsoft, Alphabet, Amazon, Broadcom, Meta, Micron and AMD all live or die on AI infrastructure spending. When people say the index is diversified across 500 companies, the weight tells a different story.

How We Got Here

Here is the S&P 500's top-ten share at the points where we could find a source we trust. The sources measure at slightly different dates, so treat each figure as roughly right, give or take a point.

When	Top 10 share	Source
End of 1985	19.5%	GW&K
1990	~19%	RBC Wealth Management
March 1999	25.2%	GW&K
2000, intra-year peak	~27%	RBC Wealth Management
2014	18.0%	GW&K
2020	28.6%	GW&K
End of 2025	~41%	RBC, FactSet data
Sept 23, 2026	40.2%	SPY holdings, Alphabet combined

For about thirty years the number lived between 18% and 27%. The dot-com mania, the most famous bubble of our lifetimes, took it to roughly 27% at the very top. In the last ten years it more than doubled, from under 20% to around 40%, and it has been parked near that level all year.

The top ten itself has turned over almost completely. In 1985 the list was DuPont, Kodak, Royal Dutch, Amoco, Shell, AT&T, General Motors, GE, Exxon and IBM. Oil, film, cars and phones. Only one company on that list, Exxon, sits anywhere near the top of the index today.

Which 40%? The Denominator Trap

You will see other charts claiming the market was just as concentrated in the 1960s. Some of them are right, and they are also measuring something different.

Morgan Stanley's Michael Mauboussin tracks the top ten as a share of the entire US stock market, thousands of companies, back to 1950. On that measure the peak was about 30% in 1963, when AT&T, GM and IBM dominated. The Nifty Fifty era in 1972 was around 15%. The dot-com top was around 19%. Early 2024 was 28%.

Those numbers run lower than the S&P figures above because the denominator is bigger. Both series are legitimate. What you cannot do is put a whole-market number from 1963 next to an S&P number from 2026 and call it a comparison, and a surprising number of viral charts do exactly that. When someone quotes a concentration figure, the first question is: percent of what?

Measured consistently, the picture holds up. On the S&P measure, today is well above every prior reading. On Mauboussin's whole-market measure, early 2024 was already within two points of the 1963 record, and concentration has kept climbing since.

1972: Great Companies, Wrong Price

The Nifty Fifty were the "one decision" stocks of the early 1970s. Buy them, never sell, because they were the best companies in America. Coca-Cola, IBM, Xerox, Polaroid, McDonald's, Avon.

Jeremy Siegel's research puts the group's average price-to-earnings ratio at 41.9 in 1972, against 18.9 for the S&P 500. Polaroid traded at 94.8 times earnings. Then 1973 and 1974 happened, one of the worst bear markets in US history, and the most expensive names were among the hardest hit.

Here is the part the doom threads leave out. Siegel found that an equal-weighted Nifty Fifty portfolio bought at the December 1972 peak still returned 12.5% a year through August 1998, nearly matching the S&P 500. The group as a whole was only modestly overpriced. The damage was concentrated in the handful of names priced for perfection, while the steady compounders eventually justified their multiples.

Lesson one: a concentrated market built on genuinely great businesses can still work out for patient holders. It can also take two brutal years to get there, and nobody trading futures has a 26-year holding period.

2000: The Lost Decade

At the March 2000 peak, Cisco was the most valuable company in the world, worth over $500 billion. It sold routers, a real product with real demand, to a real buildout that did happen. The stock price did not close above its 2000 high again until December 2025.

The index-level numbers are worse. A CFA Institute study of the 2000 to 2010 decade found:

Portfolio, 2000-2010	Annualized return
Top 500 US stocks, cap-weighted	-0.99%
Top 10, equal-weighted	-6.63%
Bottom 490, equal-weighted	+2.55%

The same study found that whenever top-ten concentration ran above 23%, the bottom 490 equal-weighted outperformed over the following five years 88% of the time. GW&K's history tells the same story from the small-cap side: after the March 2000 peak, the size premium ran about 6% a year for five years.

Lesson two: when the leaders were priced for a future that took twenty years to arrive, the index went nowhere for a decade even though the technology won.

The Case That This Time The Weight Is Earned

We would rather hand you the counter-argument than let you find it later.

These companies make enormous amounts of money. Nvidia reported $96.2 billion in quarterly revenue in August at a 75% gross margin, and guided higher. Cisco in 2000 was a fraction of that scale. The top ten today are mostly cash machines, which the 2000 leaders often were not.
The earnings gap is real but smaller than the weight suggests. At the end of 2025 the top ten held about 41% of the index's weight and produced about 32% of its earnings, per RBC. So the group carries about nine points more of the index than its share of the profits would justify. A CFA Institute study flagged a similar gap of 37% weight against 28% earnings as the widest since 1970 outside of the 2000 top, which is the bear case. Both are true at once.
Rising concentration has historically been a good time to own stocks. Mauboussin found that from 1950 to 2023 the S&P averaged 16.4% a year during periods when concentration was rising, against 10.5% when it was falling. The danger shows up when concentration starts to unwind, and it has not unwound yet.
Owning the single biggest stock used to be a losing bet. From 1950 to 2023 the number one company underperformed the S&P by an average of 1.9 points a year, per Mauboussin. That pattern flipped hard from 2014 to 2023. Whether it flips back is the whole question.

Our read: this looks more like 1972 than 2000. The leaders are real businesses with real earnings, priced at demanding multiples. 1972 still produced two brutal years. We made the fuller version of this argument in our June piece on the AI bubble question.

What It Changes If You Trade ES And NQ

This is the section that matters for your account, whatever you believe about valuations.

A handful of earnings reports are index events. Nvidia rallied 8.7% the day after its August report. At an 8% weight, that single stock moved the S&P roughly 0.7% by itself. Mark the top five report dates on the same calendar as CPI and FOMC, because for ES they carry comparable weight.
NQ is the same bet with more leverage. The Nasdaq-100 holds a fifth as many stocks and the same names sit at the top, so everything in this article applies to NQ at higher intensity. When semis sell off, NQ and ES stop behaving like different instruments.
The index can lie about the market. With 40% of the weight in ten names, the S&P can print highs while most stocks go nowhere, and it can drop while most stocks are fine. We walked through how that looks on a chart in our July index divergence piece. When YM, ES and NQ disagree, concentration is usually why.
Correlation clusters on bad days. Eight of the top ten ride the same theme. On a day when the AI trade gets questioned, you do not get 500 independent stocks averaging each other out. You get a third of the index moving together, and your stop needs to respect that.

Key Takeaways

As of September 23, 2026 the ten largest companies are 40.2% of the S&P 500. Nvidia and Apple alone are 15.6%.
For thirty years the figure lived between 18% and 27%. It roughly doubled in the last decade.
Check the denominator before trusting any historical comparison. Whole-market and S&P figures are different series.
1972 showed great companies can survive a bad entry price. 2000 showed an index can go nowhere for ten years while the technology wins.
For ES and NQ traders, mega-cap earnings are macro events and correlation spikes when the AI theme wobbles. Size for that.

Sources

IMPORTANT DISCLAIMER: This is for informational and educational purposes only and does not constitute financial, investment, trading, or tax advice. Past performance is not indicative of future results. All investments involve risk, including the potential loss of principal. Data may not be real-time or complete. Always do your own due diligence and consult a qualified financial advisor before making any investment decisions.