How Dependent Is the Stock Market on the AI Boom?
The S&P 500 is more concentrated than at the dot-com peak, with a handful of AI names carrying the index. Here is how dependent the market really is, measured with Claude.
How Dependent Is the Stock Market on the AI Boom?
How much of the market now rides on a single theme, how solid the ground beneath it is, and why you may be far more exposed to AI than you realise.
I wrote the brief, the questions and the sources. Fable 5, Claude's newest model, ran it in a single research pass.
A note on the tool: Fable 5 launched on 9 June 2026 and was withdrawn on 12 June 2026, when a US export-control directive suspended it outside the US. I ran this from London in that short window. The method matters more than the model, the same brief runs on any current Claude model.
The exposure you did not choose
If you own a plain S&P 500 index fund, roughly 35 cents of every $1 sits in just 7 companies, and somewhere between 40 and 45 cents sits in companies whose fortunes are tied to artificial intelligence. That is not a choice you made. It is what the index has become.
A shorthand for the seven mega-cap technology companies that dominate the S&P 500: Apple, Microsoft, Nvidia, Amazon, Alphabet (Google), Meta and Tesla. Together they are worth over $22 trillion and now make up roughly a third of the entire index, so when they move, the whole market tends to move with them.
This concentration is now higher than at the peak of the dot-com bubble in 2000. Unlike the dot-com era, today's giants are extraordinarily profitable, which is the strongest argument that the concentration is earned. But the boom has a circular quality: a small cluster of companies is spending roughly $700 billion this year building AI infrastructure, and much of that money flows between members of the same cluster, sometimes financed by the suppliers themselves.
None of this means a crash is coming. It means an S&P 500 tracker today is, to a meaningful degree, a bet on AI infrastructure economics. Understanding that is the point of this report.
"Concentration" just means how much of an index sits in its biggest few names. A market-cap-weighted fund like the S&P 500 does not hold 500 equal slices. The largest companies take up far more room, so when they move, the whole index moves with them, whether you wanted that bet or not.
How much of the market is the AI trade
"AI-exposed" needs a definition or the analysis cannot be audited. This note uses the Magnificent 7 (Apple, Microsoft, Nvidia, Amazon, Alphabet, Meta, Tesla) as the core, a second tier of infrastructure and enablers inside the S&P 500 (Broadcom, Micron, AMD, Oracle, Palantir, plus power and data centre names), and a broader analyst-defined cohort.
Apple is the weakest direct AI play of the 7. It is included because it is priced and traded as part of the cohort and benefits from the same passive flows. That is a judgement call, stated openly.
| Measure | Figure | As of | Source |
|---|---|---|---|
| Magnificent 7 share of S&P 500 | ~35% | Jun 2026 | Stock Analysis / Motley Fool |
| Top 10 share of S&P 500 | ~37% | Jun 2026 | SPY holdings · peak 40.7% late 2025 (RBC) |
| Top 5 share of S&P 500 | ~26% | mid-2026 | SPY holdings (SSGA) · ~30% peak late 2025 (BoE) |
| AI-linked cohort share consensus | ~45% | 2026 | Goldman Sachs |
| Nasdaq 100 top 10 weight | 46.7% | 4 Jun 2026 | Invesco QQQM |
| MSCI World top 10 weight | 27.8% | 29 May 2026 | MSCI factsheet |
The global point matters for anyone holding a "world" tracker. Of MSCI World's top 10 as of 29 May 2026, every single name is a US company in the AI cohort. A global fund does not diversify away this theme; it dilutes it slightly.
Close to $4 in every $10 in an S&P 500 tracker now sits in 10 companies. The top 10 hit a record 40.7% in late 2025 and has eased a little since, as the Magnificent 7 actually lagged the index over the first half of 2026 (up roughly 5% against the index's near-8%). Even after that cooling, this is about the most concentrated the index has been in at least 50 years, and well past the dot-com record. A global tracker softens this only modestly. Concentration alone is not a verdict; it is an exposure measurement.