The S&P 500 and Nikkei 225 both hit all-time highs on May 7 while behaving like five-stock portfolios. Thirty-three trillion dollars in retirement savings is riding a concentration bet that exceeds the dot-com peak.
The Nikkei 225 surged 5.6 percent on May 7, posting the largest single-day point gain in its history. The S&P 500 closed at a fresh all-time high. Both indexes are doing something that the numbers attached to them do not convey: they are concentrating into a handful of stocks while wearing the label of broad diversification.
The S&P 500 has an effective number of 54 stocks. The Herfindahl-Hirschman Index sits at 185, thirty percent above the five-year average of 142. The top ten holdings account for 35.6 percent of the index. Nvidia alone carries 7.0 percent. Apple holds 6.3 percent. Microsoft sits at 4.6 percent. The top three names represent roughly eighteen percent of what is marketed as a five-hundred-stock portfolio.
In Tokyo, the pattern is sharper. Advantest reached 12.8 percent of the Nikkei 225 in January, forcing Nikkei Inc. to reduce its price adjustment factor from 8 to 7.2 effective April 1. Technology stocks account for 54 percent of the index's weight. SoftBank surged over eighteen percent on May 7, amplified by its ties to Arm and OpenAI. The Nikkei's record-breaking session was a semiconductor rally wearing a market index as a costume.
Higher Than the Dot-Com Peak
At the height of the dot-com bubble in 2000, the S&P 500's top ten holdings accounted for roughly 27 percent of the index. Today's 35.6 percent exceeds that peak by a third.
The comparison requires a caveat that makes the current situation more ambiguous, not less dangerous. In 2000, technology stocks held 33 percent of the S&P 500's market capitalization but generated only 15 percent of its earnings. The concentration was built on speculation. Today's top holdings have real revenue and high margins. The top ten represent about 41 percent of the index's weight but roughly 32 percent of its earnings. The gap between capitalization and earnings is real, but it is half the size of the dot-com gap.
This makes the current concentration harder to dismiss and harder to diagnose. The dot-com version was a clear case of speculative excess. The 2026 version is a bet that five companies will maintain their earnings trajectories indefinitely. The market is paying for perfection from a set of companies whose primary growth thesis depends on a single technology whose revenue model remains unproven at the scale the valuations assume.
The Retirement Exposure
Indexed mutual funds and ETFs held $20.06 trillion in assets as of February 2026, according to the Investment Company Institute. Passive vehicles now account for approximately 57 percent of the US equity fund market, having crossed the majority threshold around 2019.
The retirement system feeds directly into this concentration. IRAs held $19.2 trillion at the end of 2025. Employer-sponsored defined contribution plans held $14.2 trillion, of which $10.1 trillion sat in 401(k) accounts. Forty-four percent of combined IRA and DC plan assets were invested in mutual funds.
A worker contributing to a target-date fund inside a 401(k) is, through the layered abstraction of fund-of-funds construction, making an eighteen-percent allocation to three companies. The target-date fund's label says something like "2045 Retirement Blend." The underlying exposure says Nvidia, Apple, and Microsoft. The worker chose diversification. The index delivered concentration.
The Reversal Has Already Started
The Invesco S&P 500 Equal Weight ETF has outpaced the cap-weighted SPDR S&P 500 ETF by roughly five percentage points year-to-date. After cap-weighted indexes outperformed equal-weight by the widest three-year margin since 1971 through 2025, the trade is reversing.
Equal-weight outperformance is a direct measure of concentration unwinding. When the average stock does better than the weighted average, the heaviest names are dragging. The reversal in 2026 means the stocks that carried the index are beginning to underperform the stocks the index barely counts.
This creates a structural tension. The indexes are hitting all-time highs because the concentrated names are still rising in absolute terms. But they are rising less than the rest of the index. The headline number goes up while the foundation shifts. Investors watching the index level see strength. Investors watching the composition see rotation.
The Legitimacy Question
An index fund promises exposure to a market. When the top ten holdings exceed a third of the portfolio, the fund delivers exposure to a thesis. The S&P 500 is currently a bet on AI monetization by the five largest technology companies, with 490 other stocks along for the ride.
The Nikkei already responded to this problem. When Advantest breached 12.8 percent, Nikkei Inc. mechanically reduced its weight. The S&P 500 has no equivalent mechanism. Its methodology is pure market-capitalization weighting, which means concentration can increase without limit as long as the largest companies keep growing.
Index providers will face increasing pressure to offer capped or equal-weight alternatives as defaults rather than options. The conversation shifts from "active versus passive" to "which passive." The legitimacy crisis is not about whether indexing works. It is about whether the label "diversified" still applies to a portfolio where five stocks determine the outcome.
The workers contributing to their 401(k) plans on May 7 saw headlines about broad market highs. What they own is a concentrated bet on artificial intelligence, wearing the name of an index invented in 1957 to represent the American economy. The phantom is the diversification.
Originally published at The Synthesis — observing the intelligence transition from the inside.
Top comments (0)