7 ways to diversify away from AI stocks
You Own the AI Bet. Now What?
I will walk you through what an ordinary investor can actually do to diversify away from AI stocks: every option available to reduce the exposure, how each one works, and what each one costs. But the honest answer is not a fix, it is a personal choice.
The bet you already hold
The reliance report measured the bet. This one starts from its conclusion and does not re-argue it: in a standard S&P 500 tracker, of every $1 you hold, roughly 35 cents sits in the Magnificent 7 and about 45 cents sits in the wider AI-linked cohort. That is the same scope and the same numbers as last week, carried straight over.
Start here Read "How Dependent Is the Stock Market on the AI Boom?" first →So the question is no longer how big the bet is. It is what, if anything, you do about it. And here the instinct, to reach for the thing that "fixes" the concentration, runs straight into an uncomfortable truth: nothing on the list reduces risk outright. Every option swaps one risk for another. You give up concentration risk and take on tracking-error risk, the risk of looking wrong against the headline index for years at a time.
De-concentrating is risk substitution, not risk reduction. The question is never "how do I make this safer." It is "which risk would I rather hold, and can I stick with it when it looks like the wrong call."
Measure what you actually own
Most people do not have one AI bet. They have the same one stacked 3 or 4 times and do not see it. An S&P 500 tracker, a Nasdaq 100 fund, a "global" tracker that is mostly the United States, a tech ETF, and most actively managed US funds all carry the same handful of names at the very top. Someone who feels diversified across 5 funds can easily be a third to a half of everything in the same 7 companies.
The look-through, in 3 steps
This is the exact move a professional allocator makes before worrying about anything else. It is called look-through analysis, and you can do it in an afternoon:
For every fund you hold, find its top 10 holdings and their weights. Every fund publishes this on its factsheet.
Mark which of those names repeat across your funds. The Magnificent 7 and the wider AI cohort will keep reappearing.
Weight each fund by how much of your money sits in it, then add up the repeated names. That total is your true exposure.
Say a portfolio is 50 percent in an S&P 500 tracker, 30 percent in a global tracker, and 20 percent in a Nasdaq 100 fund, and the holder feels well spread across 3 funds. Blend the Magnificent 7 weights, roughly 35 percent, 22 percent and 56 percent, by those holdings and you land at about 35 percent of the whole portfolio in 7 companies. Add the wider AI-linked cohort and it climbs well beyond that. 3 funds, one bet. The figures are illustrative; your own look-through is the one that matters.
You cannot manage a bet you have not measured. Before any decision about reducing exposure, find your real number across every fund at once. It is almost always higher than any single fund's headline, and higher than it feels.
Reweight the same market
3 of the 7 options work by changing how the index is weighted. They all cut how much of your money sits in the biggest names, but they keep you largely in the same US companies. So they reduce the concentration more than they add a genuinely new driver. Funds are named only to show how each one works in practice, never as a recommendation to buy or sell.
You hold the same 500 companies, but instead of weighting by size, where the biggest names get the most money, every company gets roughly the same slice, about 0.2 percent each (100 percent split 500 ways), rebalanced back to equal every quarter. The Magnificent 7 fall from about 35 percent of the fund to roughly 1.4 percent combined, that is 7 names at about 0.2 percent each.
The ready-made version is an equal-weight S&P 500 fund. In the US the original is the Invesco S&P 500 Equal Weight ETF (RSP); UK and European investors have UCITS equivalents. You can hold it instead of a standard tracker, or, more commonly, alongside one, so part of your money follows each rule. For example, holding the equal-weight version in place of a standard tracker takes the Magnificent 7 from about a third of your money to under 2 percent combined.
It trailed the standard S&P 500 through the mega-cap run, and the gap was wide. Over the 10 years to June 2026 the cap-weighted S&P 500 returned about 320 percent in total against about 208 percent for the equal-weight version; over 5 years, about 90 percent against about 55 percent. It also tilts you toward mid-caps and value and charges a little more, about 0.20 percent versus under 0.10 percent. It needs market breadth to win. See the chart below.
A gentler version of equal weight. Rather than flattening everything, you keep the size-based order but put a ceiling on any single name, or balance only within the largest tier. The giants get trimmed; the rest of the shape stays roughly the same.
2 common forms. First, capped index funds: many UCITS funds already apply a 10 percent single-name cap, and capped versions exist that hold no name above a set limit. Second, large-cap equal weight: a fund that equal-weights only the 100 largest companies (an equal-weight S&P 100, such as EQWL) keeps you in mega-caps but stops any one of them dominating. For example, switching a standard tracker for a large-cap equal-weight fund keeps you in the giants but stops any one of them dominating.
The cost here has been small, because you still hold the giants. Over the 10 years to June 2026 the large-cap equal-weight S&P 100 returned about 300 percent in total against about 320 percent for the standard S&P 500; over 5 years, about 88 percent against about 90 percent, effectively level. You give up very little long-run return, but you also keep more of the concentration than full equal weight, so your exposure falls by less. It dilutes the bet rather than removing it.