About 38 cents of every dollar in an S&P 500 index fund now sits in just ten companies, and most of them are riding the same AI spending wave. That alone isn’t a reason to sell anything. The real concentration risk usually isn’t the index fund. It’s everything you stacked on top of it.
The index didn’t change. The market did.
An S&P 500 fund doesn’t pick winners. It owns every company in proportion to its size. When a handful of companies get enormous, the fund gets concentrated automatically, no decision required.
As of June 30, the ten largest holdings made up 37.9% of Vanguard’s S&P 500 ETF [2]. That list is NVIDIA, Apple, Alphabet, Microsoft, Amazon, Broadcom, Micron, Meta, Tesla, and Eli Lilly. By mid-2025, the top ten had reached almost 40% of the index, a level not seen since the mid-1960s [1].
That sounds alarming until you remember how the machine works. If those ten companies fall, their weight falls with them. Nobody at the fund company has to sell anything, and neither do you. Cap weighting is self-correcting in a way no stock picker is.
It also cuts the other way. If the AI giants keep winning, you keep owning more of them. The index never argues with the market. It just holds it.
The industry’s standard answer to concentration is a product: an equal-weight fund, a “diversified AI” fund, a managed account with a quarterly fee. A few of those have a real case. Even the plain version costs more. A popular equal-weight S&P 500 ETF charges 0.20% a year [VERIFY: confirm RSP expense ratio against current Invesco prospectus] [5]. Vanguard’s cap-weighted S&P 500 ETF charges 0.03% [2]. On $25,000, that’s $50 a year versus $7.50, every year, compounding against you.
You probably don’t need a new product. You need an inventory.
Where the real concentration hides
Picture a portfolio that looks perfectly sensible, one account at a time.
- 401(k): an S&P 500 index fund, because it was the cheapest option on the menu.
- Roth IRA: a total U.S. market fund, because that’s the textbook recommendation.
- Taxable brokerage: a technology sector ETF, bought last year “for a little AI exposure.”
- ESPP: some company stock, because the 15% discount felt like free money, and the employer is one of the giants.
Every one of those choices is defensible. Together, they’re the same bet made four times.
Run the overlap math
Take a $50,000 portfolio, right in Marcus territory. The weights below come from each fund’s June 30, 2026 fact sheet [2][3][4]. For the tech fund, count only its stakes in the same ten companies that top the S&P 500.
| Holding | Balance | Weight in the top-10 companies | Dollars in the top 10 |
|---|---|---|---|
| S&P 500 ETF (401k) | $25,000 | 37.9% | $9,475 |
| Total U.S. market ETF (Roth) | $15,000 | 33.4% | $5,010 |
| Info tech ETF (taxable) | $6,000 | 47.9% | $2,874 |
| Employer stock (ESPP) | $4,000 | 100% | $4,000 |
| Total | $50,000 | $21,359 |
That’s 42.7% of the whole portfolio sitting in ten companies.
Now put the same $50,000 in a plain three-fund portfolio: 60% total U.S. market, 30% total international, 10% bonds. Only the U.S. slice holds those ten companies. $30,000 times 33.4% is $10,020, or 20.0% of the portfolio.
Same net worth. More than double the bet.
Nobody sat down and decided to put 43% of their money into ten stocks. It happened one reasonable purchase at a time. That’s the version of concentration risk worth worrying about, because it’s the one you didn’t choose.
What a sensible amount looks like
Your baseline is whatever the broad market hands you. U.S. companies are 63.6% of MSCI’s global stock index, which leaves roughly 36% of the world’s market outside the U.S. [6]. Add international and your top-10 share drops without you betting against anything.
If your top-10 exposure lands well above what a plain total-market-plus-international mix would give you, you’ve made an active bet. Active bets are allowed. They should just be on purpose, and sized so that a bad few years for AI stocks doesn’t wreck your plan for the next twenty.
FINRA flags two ways investors get here without meaning to: loading up on employer stock, and holding different investments that move together. Its fix is the same one this piece runs, which is to look inside every fund you own for overlapping holdings [7]. Company stock deserves the hardest look. Your paycheck already depends on that employer, so a bad year for the company can hit your job and your savings at the same time.
And the top ten could keep outperforming for years. Trimming the overlap can cost you returns in that world. You’re not predicting a crash. You’re deciding how much of your future you want riding on one theme.
Fix it without a tax bill
The instinct is to sell something. In a taxable account, selling a winner triggers capital gains tax. Gains on assets held more than a year are long-term and may qualify for a lower rate. Gains on assets held a year or less are short-term [8]. Inside a 401(k) or Roth, you can swap funds with no tax at all. That difference decides the order you fix things in.
Run the same $50,000 through three moves. Ignore market growth, to keep the math honest.
Step 1: Sell the ESPP shares on a schedule and put the proceeds in total international. ESPP shares get the best tax treatment only if you hold them more than two years from the grant date and more than one year from the purchase date. Sell earlier and part of the gain can be taxed as ordinary income [9]. Check your Form 3922 dates first. After the sale, top-10 dollars drop to $17,359, which is 34.7%.
Step 2: Point the next $12,000 of contributions at the boring funds. Two-thirds goes to total U.S. market and one-third to international. That adds $8,000 × 33.4% = $2,672 of top-10 exposure. The total is $20,031 on a $62,000 portfolio, or 32.3%.
Step 3: Inside the 401(k), switch from the S&P 500 fund to total market plus international, if your plan offers them. Split the $25,000 into $17,500 U.S. and $7,500 international. The U.S. piece carries $5,845 of top-10 exposure instead of $9,475. The total falls to $16,401, or 26.5%.
From 42.7% to 26.5%. No taxable sale beyond the ESPP, and you still own every one of those AI companies, just in proportion to their actual share of the market.
The tech ETF can stay. Stop feeding it.
The move to make
Tonight, open every account and write down four things for each fund: its name, the dollar balance, the top-10 weight from its fact sheet, and the dollars that works out to. Add up the dollars and divide by your total portfolio.
Then run the same math on a plain total-market-plus-international mix at your stock allocation. For the portfolio above, that benchmark is about 20%. If your real number runs more than ten points higher, you’re carrying a bet you didn’t choose. Fix it in this order:
- Do the tax-free swaps inside your 401(k) and Roth first.
- Redirect new contributions to total market and international.
- Sell taxable positions last, and only when the tax cost is small.
Don’t sell your index fund. Just stop stacking copies of it on top of itself.
Sources
Last reviewed: September 29, 2026. Fund weights change daily; figures are as of each document’s date.
- S&P Global, “Partner Perspectives: Unlocking Potential Ahead.” The 10 largest S&P 500 companies represented almost 40% of the index by mid-2025, a level not seen since the mid-1960s. spglobal.com
- Vanguard S&P 500 ETF (VOO) fact sheet, as of June 30, 2026. Expense ratio 0.03%; ten largest holdings NVIDIA 7.5%, Apple 6.6%, Alphabet 5.8%, Microsoft 4.3%, Amazon 3.6%, Broadcom 2.8%, Micron 2.0%, Meta 1.9%, Tesla 1.8%, Eli Lilly 1.5%; top ten 37.9% of total net assets. fund-docs.vanguard.com/F0968.pdf
- Vanguard Total Stock Market ETF (VTI) fact sheet, as of June 30, 2026. Same ten largest holdings as VOO; top ten 33.4% of total net assets. fund-docs.vanguard.com/F0970.pdf
- Vanguard Information Technology ETF (VGT) fact sheet, as of June 30, 2026. NVIDIA 16.2%, Apple 14.4%, Microsoft 8.3%, Micron 5.1%, Broadcom 3.9%; these five overlap the S&P 500’s top ten, for a combined 47.9%. fund-docs.vanguard.com/F0958.pdf
- Invesco S&P 500 Equal Weight ETF (RSP), expense ratio. Net expense ratio 0.20% per Yahoo Finance; pending confirmation against the Invesco prospectus. finance.yahoo.com/quote/RSP
- MSCI ACWI Index factsheet, as of August 31, 2026. Country weights: United States 63.59%. msci.com
- FINRA, “Concentrate on Concentration Risk.” Identifies company-stock concentration and correlated-asset concentration as risks, and advises checking each fund’s prospectus or website for overlapping holdings. finra.org
- IRS Topic No. 409, Capital gains and losses. Assets held more than one year produce long-term gains or losses; one year or less, short-term; a lower rate may apply to net capital gain. irs.gov/taxtopics/tc409
- IRS guidance on employee stock purchase plans. Holding period runs 1 year after transfer and 2 years after grant; Form 3922 tracks both; failing the holding period can still produce ordinary income. irs.gov/node/17603 , irs.gov/node/17564
The post Index fund concentration risk: how big an AI bet you’re actually making appeared first on Thousandaire.
