BUSINESS
The S&P 500 No Longer Shelters Investors From AI
Index funds now fail a 1940 diversification test as 10 stocks hit 38%, and equal-weight shelters still hold the same 500 names.
The S&P 500’s 10 largest stocks were 38% of the index in August 2026, enough to collide with a 1940 legal test of a diversified fund. Nvidia is 8.09% of Vanguard’s S&P 500 ETF and Apple is 7.04%, so two names now outweigh hundreds of others in the same product.
Investors are being offered equal-weight funds, gold, dividend screens and smaller equity sleeves as shelter. Those products still leave most of the stock sleeve inside the same 500 companies, and the cheap cap-weighted fund still collects the retirement money.
The Default 401(k) Is Now an AI Fund
A paycheck that lands in a target-date or S&P 500 fund does not spread itself across 500 equal bets. It follows market value. As the AI buildout lifted a handful of platforms, the default U.S. equity holding became a concentrated wager on the same spending cycle.
Holdings feeds for Vanguard’s S&P 500 ETF in early October listed Nvidia at 8.09%, Apple at 7.04% and Microsoft at 5.70%. Alphabet’s two share classes add to 5.41% (3.01% in the A shares and 2.40% in the C shares). Those four issuers sum to 26.24% of the fund, above the 25% line that sits inside the 1940 Act diversified-company test.
The Investment Company Institute, using Morningstar Direct and S&P Global data, said the top 10 holdings constituted 38% of the S&P 500 in August 2026. Four S&P names and five Russell 1000 Growth names each stood above 5%. The Russell 1000 Growth’s top 10 were 57% of that index.
HOW THE TOP OF THE FUND IS LOADED
- Nvidia: 8.09% of Vanguard’s S&P 500 ETF.
- Apple plus Nvidia: 15.13% in two stocks.
- Four issuers above 5%: 26.24% combined, once Alphabet’s classes are counted as one company.
- Top 10 of the S&P 500: 38% of the index in August 2026.
The label on the statement still says 500 stocks. The math says a bad week in chips or phones now moves the whole sleeve.
Equal Weight Still Owns the Same 500 Stocks
The product sold as the clean fix is the Invesco equal-weight S&P 500 ETF, ticker RSP, launched April 24, 2003. It holds the same index members and resets them to equal weights each quarter, about 0.20% each. On Oct. 2, 2026, it had 506 holdings and $95.43 billion in market value. The fee is 0.20%.
That reset cuts the Nvidia-size bets. It does not leave the AI trade. On Oct. 2 the largest lines were still software, tools and chip names such as Moderna, Illumina, Crowdstrike, Synopsys, Cadence, Datadog and Advanced Micro Devices, each around 0.29% to 0.25%. Information technology was 15.67% of the fund on Sept. 30, 2026, with industrials at 16.60% and financials at 14.67%. You own more of the smaller 500, including the ones tied to the same buildout.
Through Aug. 31, 2026, that mix was finally ahead of the cap-weighted index for the year. It was still behind over every longer window Invesco publishes against the S&P 500.
THE TWO WAYS TO OWN THE S&P 500
| Measure (through Aug. 31, 2026, unless noted) | S&P 500 (cap-weighted) | RSP (equal-weight) |
|---|---|---|
| Expense ratio | 0.03% (Vanguard S&P 500 ETF) | 0.20% |
| Year-to-date return | 13.14% | 15.44% (NAV) |
| 1-year | 20.38% | 18.25% |
| 3-year annualized | 21.04% | 15.43% |
| 5-year annualized | 12.79% | 8.80% |
| 10-year annualized | 15.37% | 12.00% |
| Largest line | Nvidia 8.09% (VOO, early October) | about 0.29% |
The five-year gap is 3.99 percentage points a year in favor of the top-heavy index. The 10-year gap is 3.37 points. Since inception the two paths are close, 11.65% a year for the S&P 500 versus 11.39% for RSP. The shelter has a bill, and until 2026 most holders refused to pay it.
The Cheap Index Still Collects the Retirement Flows
Vanguard’s S&P 500 ETF charged 0.03% and held $979 billion on June 30, 2026; the broader 500 Index fund complex sat at $1.675 trillion. RSP’s 0.20% fee is 0.17 percentage points higher. On a $100,000 sleeve that is $170 a year before any return gap.
That price gap, plus years of watching Nvidia pull the cap-weighted index higher, is why the “broad” fund remains the default. Equal-weight’s 2026 lead through August does not erase a 3.99-point five-year shortfall. Plan sponsors who stayed in the 0.03% share class were paid to stay concentrated.
The index can print a gain while most members lag, because a few mega-caps carry the average. That is not a secret on trading desks. It is the reason a 401(k) that never bought an AI fund still lives on AI earnings. Stepping into RSP is a bet that the other 490 names will matter more from here, not a way to stop owning them.
The Last Time 10 Names Ran the Index
High concentration is not new. The ICI note is blunt that the S&P 500 was similarly top-heavy in the late 1960s and early 1970s. What changed is the wrapper count. There were 361 mutual funds in 1970. At year-end 2025 there were 11,264 mutual funds and ETFs. A legal test written for a smaller industry now hits almost every cap-weighted core fund at once.
THE LEGAL AND MARKET CLOCK
- Late 1960s to early 1970s: The S&P 500’s top 10 reach weights the ICI calls similar to today’s.
- August 2026: The top 10 are 38% of the S&P 500; four issuers each exceed 5%.
- September 10, 2026: The ICI asks the SEC for faster relief so funds are not trapped between tracking error and proxy fights.
- September 29, 2026: Fidelity’s board backs a shareholder vote to reclassify the Fidelity SAI U.S. Momentum Index Fund as non-diversified.
- First quarter of 2027: That Fidelity meeting is scheduled; if holders approve, the change is set for April 1, 2027.
A fund that tracks a top-heavy index cannot both hug the benchmark and stay inside the old 5% and 25% arithmetic. Vanguard has already told 500 Index shareholders to watch “nondiversification risk,” language that appears when market moves, not stock picks, push a tracker over the line.
Funds Are Asking Voters to Drop “Diversified”
Under the 1940 Act, a fund that calls itself diversified must keep at least 75% of assets in a basket where no single issuer is more than 5% of the fund (and no more than 10% of that issuer’s voting stock). The other 25% has more room. When several stocks each clear 5% and those positions sum to more than 25%, the label breaks.
Index funds got a partial pass in 2019, when SEC staff said it would not recommend enforcement if a broad-index tracker kept buying a name that had grown through the cap. Active funds did not get the same letter. Changing a fundamental policy still needs a shareholder vote.
THE PROXY BILL, 2020 TO 2025
- 73 funds: Sought to drop diversified status, per the ICI’s 2026 white paper.
- $66 million: Combined campaign cost, over $900,000 on average; three campaigns each topped $5 million.
- 85% yes: Average support among shares represented, above the 67% statutory mark; 92% if abstentions and broker non-votes are stripped out.
- Six failures: Died for lack of quorum, even with that support.
Holders who bother to vote want the fund to follow the index. The process is slow and expensive, which is why a backlog remains. Fidelity’s September 29, 2026, supplement is the live case: the momentum index fund would otherwise have to cap winners or miss its benchmark.
Where the Money Goes When AI Wobbles
BlackRock is not telling clients to sell the theme. Its fall 2026 note says the firm remains convicted in AI equities while diversifying around AI portfolio concentration. Q2 earnings are the exhibit. AI infrastructure companies grew earnings 54% from a year earlier; the rest of the S&P 500, as BlackRock screens it, grew 14%. Eight of 11 sectors posted double-digit earnings growth.
Evidence from Q2 earnings showed that the AI opportunity is becoming broader, not more zero-sum. We now see investable opportunities in, around and beyond the theme.
Gargi Pal Chaudhuri, Chief Investment and Portfolio Strategist, Americas, BlackRock
The summer still left a mark. From June 22 to July 29, dividend-paying stocks gained 8% during an AI sell-off. The PHLX Semiconductor Index moved more than 5% up or down on 33% of trading days in June and July, versus 3% of days over the prior 10 years. Average risk on the moderate advisor portfolios BlackRock screens rose to 11.3% from 10.8% at the start of 2026.
The same five hyperscalers, Amazon, Microsoft, Alphabet, Meta and Oracle, issued about $200 billion of investment-grade debt in the first half of 2026, nearly double their issuance in all of 2025. That is how the trade leaks into the bond sleeve that is supposed to cushion it. Bond mutual funds and ETFs still took about $912 billion in 2026, a caution trade running beside the equity rally.
Gold, Dividends and a Smaller Equity Sleeve
AllianceBernstein’s Oct. 2, 2026, outlook said global stocks in the MSCI ACWI rose 1.6% in the third quarter as leadership broadened, with the MSCI ACWI Value Index up 3.8% in the quarter and 16.1% on the year. Tech volatility, on AB’s screen, reached its highest since the early-2000s bust. Global government bond yields sat near 4%, the most since 2007, after a 25 basis-point Federal Reserve hike in mid-September.
AB’s conclusion is the hard one for anyone who thinks a second U.S. equity fund is enough. AI spending is tying together stocks, credit and even parts of emerging markets that used to move on their own, which makes it harder to achieve diversification with the old sector-and-style toolkit. European and Japanese stocks, some emerging markets outside the chip cohort, energy, healthcare and smaller companies that can turn AI into cash rather than spend it are the sleeves AB flags as less tied to the U.S. mega-cap cluster.
Jeffrey Gundlach, founder of DoubleLine, cut a recommended equity weight from 40% to 30% and parked that stock sleeve in an equal-weighted Fortune 500 book. He told the Julia La Roche Podcast, “I want out,” and warned there will be “fallout and losers in the AI race for the holy grail.” He also pointed to a Shiller CAPE ratio above 42.
GUNDLACH’S REPLACEMENT MIX
- 30% equities: Equal-weighted Fortune 500, the anti-cap-weight choice.
- 30% bonds: Split between high-quality paper and local-currency emerging-market debt.
- 10% gold and 10% commodities: The non-stock ballast.
- 20% dry powder: Commercial real-estate debt and a flexible income sleeve.
That mix is a full exit from the idea that 500 U.S. names, equally or not, are a complete portfolio. Most 401(k) menus will not look like it. They will still offer the 0.03% S&P fund, maybe an equal-weight cousin, and a bond option whose largest issuers now include the same hyperscalers.
The Fidelity vote in the first quarter of 2027 will not shrink Nvidia’s weight. It will only let one more fund admit what the index already is. Until the other 490 companies take the earnings baton, sheltering from AI concentration means paying 0.17 points more, owning the same list in different sizes, or leaving U.S. large-cap funds altogether.
Disclaimer: This article is news reporting and analysis for information only. It is not investment advice, a recommendation to buy or sell any fund, stock or bond, or a suggestion to change a retirement allocation. Readers should consult a licensed financial advisor or registered investment adviser who can review their own holdings, time horizon and tax situation before acting. Figures and fund statuses come from the company, index and regulator sources cited with their dates and can change with markets, flows and shareholder votes.
-
ENTERTAINMENT2 months agoRolex Made Drake the Daytona It Fights Jewelers Over
-
BUSINESS2 months agoTreasury’s First Iran Bank Shot Lands on an Ally
-
NEWS2 months agoInstinct’s $2.5 Billion Raise Still Binds the User as Agent
-
NEWS2 months agoCity’s £125m Enzo Deal Caps a £327m Midfield Rebuild
-
NEWS2 months agoCSIRTs Inherit Europe’s Missing security.txt Before Article 14
-
BUSINESS2 months agoBerkshire Anchors Alphabet’s Record Raise With a $10 Billion Check
-
NEWS2 months agoMeta’s Teen Settlement Leaves Chat Off the Clock
-
ENTERTAINMENT2 months agoRicky Gervais Wants Alley Cats to Run Until He Dies
