How concentrated is "concentrated," really
The numbers behind this worry are not vague. Ten years ago, the five largest companies in the S&P 500 were worth roughly as much as the bottom 225 stocks combined. Today, the top five match the bottom 434 — essentially the rest of the index. The ten largest holdings now make up around 40% of the index’s total value, up from a range of roughly 20-25% that held fairly steady from 1980 through the 2010s. To find a comparable concentration level, you’d need to go back to the 1950s and 1960s.
This matters for diversification because a standard S&P 500 index fund is market-cap weighted by design: the bigger a company gets, the more of your money is automatically allocated to it. That’s not a flaw introduced by AI euphoria — it’s how cap-weighted indexing has always worked. What’s changed is how much weight a small cluster of companies can accumulate before the mechanism starts to feel uncomfortable.
It’s worth separating the sentiment from the fundamentals here. The ten biggest S&P 500 companies account for about 40% of the index’s market cap, but also roughly 40% of its earnings. In other words, the market isn’t entirely detached from profitability — the largest companies are, for now, also the most profitable ones. That doesn’t mean the concentration is risk-free; it means the story is more complicated than "a bubble inflated by hype alone."
Diversification as a response to uncertainty, not a prediction
It’s tempting to treat the decision to diversify away from AI-heavy mega-caps as a prediction that those stocks are about to fall. That framing is a trap. Nobody can reliably say whether AI capital spending keeps accelerating, plateaus, or reverses, and nobody can say whether today’s concentration is a temporary phase of this bull market or a durable new structure of the economy. Diversification doesn’t require an answer to either question.
The more useful way to think about it: diversification is what you do because you don’t know the answer, not because you’re confident the answer is bad for AI stocks. An investor who sells everything related to AI on the theory that concentration "must" reverse is making a forecast just as specific — and just as unverifiable — as the investor who is all-in on the Magnificent Seven. Both are concentrated bets; they just point in opposite directions.
The trap: swapping one concentrated bet for another
This is the part that gets lost in a lot of concentration anxiety. If you dump your index fund and rotate heavily into, say, one sector ETF, one country, or one theme that’s "not tech," you haven’t actually diversified — you’ve just chosen a different single bet and are now hoping it offsets the first one. True diversification spreads exposure across multiple independent sources of return, not just away from whatever is currently popular.
A genuinely diversified response usually involves several moving pieces at once: some mix of investment style (value, quality, dividend-paying, low-volatility), sector balance, geography, and asset class (bonds, real estate, cash). Each of these addresses a different piece of concentration risk, and each leaves a different piece untouched.
| Diversifier | What it reduces | What it does NOT protect against |
|---|---|---|
| Broad market-cap index fund | Single-stock and company-specific risk | Concentration risk from mega-cap weighting itself |
| Ex-mega-cap / equal-weight tilt | Dependence on a handful of giant companies | A broad market downturn affecting most stocks |
| Dividend, value, or quality tilts | Heavy tech/growth sector weighting | Equity market risk generally; still stocks |
| Low-volatility strategies | Drawdown severity in equity selloffs | Opportunity cost if leadership stays with growth |
| Bonds | Correlation with equity market swings | Inflation risk; rate-driven bond losses |
| Foreign developed-market stocks | U.S. and dollar concentration | Currency and country-specific risks |
| REITs | Sector overlap with tech-heavy indexes | Interest-rate sensitivity of real estate |
None of these is a complete answer by itself. Each trades one exposure for a different one — which is the whole point of combining several of them rather than picking a single "best" alternative.
What the last concentrated era suggests — with real caveats
History offers a hint, not a rule. During the 2000s, when the stock market fell sharply twice, concentration in the largest stocks also fell sharply — leadership rotated toward sectors and styles that had been left behind in the prior decade. That episode is often cited as evidence that diversified, fundamentally driven strategies can regain relevance once a concentrated bull market ends.
But it is only one episode, and applying it mechanically to today’s AI-driven market is a stretch. Rate environments were different, the composition of the largest companies was different, and there’s no way to know whether a future downturn triggered by an AI spending slowdown would unfold the same way. If AI capital expenditure did cool, it’s plausible that interest rates would fall, the dollar would weaken, and non-U.S. and value-oriented stocks would hold up relatively better — but "plausible" is not the same as "guaranteed," and these pieces don’t have to move together.
Weighing the tradeoff: tracking error versus concentration risk
flowchart LR A[Notice AI concentration] --> B[Stay market-cap weighted] A --> C[Tilt within equities] A --> D[Add other asset classes] C --> E[Accept tracking error vs index] D --> F[Accept lower expected growth link to AI]
Every branch carries a cost. Staying fully indexed means accepting that your returns are increasingly tied to a small group of companies — but you also keep full participation if those companies keep growing. Tilting toward value, dividends, or ex-mega-cap strategies reduces that dependence, but it introduces tracking error: periods, sometimes long ones, where you visibly underperform the index you used to hold. Adding bonds, REITs, or foreign stocks diversifies the portfolio’s response to a shock, but it also means giving up some of the upside if AI-driven growth simply continues.
There is no setting on this dial that removes risk entirely. A globally diversified, multi-asset portfolio can reduce single-stock and sector risk, but it will not be immune to a broad market selloff — stocks, bonds, and real estate have all fallen together in past crises.
The practical takeaway
The useful response to AI concentration isn’t a prediction about where AI stocks go next. It’s a structural one: spread exposure across styles, sectors, and asset classes deliberately, understand exactly what each piece gives up in exchange for the risk it reduces, and resist the urge to replace a feared concentrated bet with an opposite concentrated bet dressed up as caution. Diversification won’t guarantee you avoid losses, and it won’t tell you whether today’s dominant companies keep dominating. What it can do is make sure that if the story changes, your entire financial future doesn’t hinge on guessing when.
