The index stopped being a broad measure#
Concentration like this is unprecedented. The top ten weighting hit a record 40.7% in 2025 and has stayed there, roughly 50% more concentrated than at the height of the 2000 tech bubble.
That changes what the number means. An index designed to represent 500 companies is now, mathematically, a bet on about ten — with two names alone accounting for something like 7% each.
One fair caveat before the pessimism: today’s giants are not 1999’s. They’re enormously profitable, generating real free cash flow, with entrenched positions. Valuation multiples aren’t as unhinged as they were in 2000. The risk isn’t that these are fake companies — it’s structural. When four-tenths of the index moves together, diversification you thought you had isn’t there.
The gap between the index and the household#
Consumer sentiment recently hit a record low in a survey running 75 years — lower than during COVID, lower than 2008 — while the index printed record highs. Those two facts sit in the same economy.
The mechanism is ownership. The wealthiest 10% of Americans hold about 93% of US equities; the top 1% around 53%; the bottom half roughly 1%. The median stock-owning family holds about $52,000 including retirement accounts, while top-decile brokerage balances run past $1 million.
So the index rising is a real event that transfers real wealth, to about a tenth of households. Everyone else reads about a boom they have no position in.
What that produces is the k-shaped split: the top 20% of households, earning roughly $175,000 and up, now account for close to 60% of all US consumer spending, with their spending growing around 6.5% year on year against 2.6% for everyone else — the latter losing to inflation.
The AI spending is now large enough to distort the data#
This is the part that deserves more attention than the stock prices.
Big tech capital expenditure — Amazon, Microsoft, Alphabet, Meta, Oracle — is estimated at $800 billion to $1 trillion this year. That exceeds the entire GDP of Sweden, around $760 billion. Five companies.
Spending at that scale doesn’t just move share prices, it moves national accounts. Analysts estimate AI-related capex drove roughly three-quarters of Q1 US GDP growth; strip it out and growth falls to about half a percent. The economy isn’t merely correlated with the AI trade at this point — a large part of measured growth is the AI trade.
And the same spending is displacing hiring. Budget shifting from headcount to infrastructure shows up as entry-level hiring down 6% year on year, which is the mechanism by which a boom coexists with graduates unable to find work.
The number I’d watch most: the gap between AI capital spending and AI revenue is around 46%, wider than the roughly 32% gap at the dot-com peak, per Allianz Research. That’s not proof of a bubble — infrastructure legitimately precedes revenue. But it is a large bet that revenue arrives on schedule, and the last time the gap was smaller, it didn’t.
Where the stress is showing#
Beneath the index: real average hourly earnings down 0.3% over the past year, the personal savings rate under 3%, household and auto debt at record highs, and auto loan delinquencies the highest ever recorded — worst among 18–29 year-olds, whose delinquency rate has roughly doubled year on year.
That last figure is the one I’d take most seriously. Young borrowers defaulting on cars at record rates is not a lagging indicator of anything good.
The honest conclusion#
The uncomfortable part is that the diagnosis doesn’t produce a clever action. You cannot fix concentration, or the ownership distribution, or where GDP growth is coming from.
What follows is narrower and duller. Since this system routes gains to owners of assets rather than earners of wages, the individual lever is to become an owner earlier and more consistently than feels comfortable — index funds, tax-advantaged accounts, automatically.
With the caveat that matters: emergency fund first. In an economy where the middle is treading water and one bad quarter from cutting back, being a forced seller is the specific failure to avoid. Getting invested early only works if you’re never made to sell early.
And hold the concentration figure in mind when you tell yourself an index fund is diversified. It’s diversified compared to owning one stock. Against 40.8% in ten names, it’s less diversified than the word implies — which is an argument for international exposure, not for abandoning the approach.
So what to actually do#
- Check the top-ten weighting of any index fund you hold. You are more concentrated than you think.
- Add international exposure if your portfolio is all US large-cap.
- Stop reading index highs as economic news. They’re a report on ten companies’ expected profits.
- Emergency fund before investing, without exception, in this environment.
- Automate ownership anyway. The distribution is what it is; being outside it is worse than being late to it.
Sources & further reading#
- Top 10 stocks account for nearly 40% of the S&P 500 — Pensions & Investments — the concentration figure and its historical comparison.
- How extreme is market concentration? — J.P. Morgan Asset Management — the 40.8% versus 26.6% dot-com comparison and why today’s leaders differ from 2000’s.
- The ‘Great Narrowing’: S&P 500 concentration — RBC Wealth Management — analysis of what concentration does to index behaviour.
- The wealthiest 10% of Americans own 93% of stocks — Yahoo Finance on Federal Reserve data — the ownership distribution figures.
