June 23, 2026

The Concentration You Cannot See

Why a well-diversified portfolio may be a far bigger bet on artificial intelligence than it appears.

Diversification is often called the only free lunch in investing. Spread your money across enough different things, the idea goes, and the disappointments in one place are offset by gains in another. It is sound, and it is the right place to start. But it rests on an assumption that is quietly breaking down: that the things you own are genuinely different from one another. Increasingly, they are not.

A large and growing share of a typical portfolio now depends, underneath, on the same single outcome: the success of the build-out of artificial intelligence. A portfolio can look well spread, across shares, bonds and other assets, across regions and sectors, and still be leaning, beneath the surface, on one theme. Here is how that has happened.

The scale of the build-out

It is easy to underestimate the scale. The handful of very large technology companies driving the build-out, the so-called hyperscalers, spent on the order of $400 billion between them on capital projects in 2025, chiefly data centres, the chips inside them, and the power to run them. Current guidance points to that rising towards $600 billion and beyond in 2026. This is investment on the scale of a national infrastructure programme, undertaken by a small number of firms, in just a few years. Spending of that size does not stay neatly inside the technology sector. It reaches into the markets that fund it, the commodities and power it consumes, and the wider economy it stimulates.

The shares you own

Start with shares, where the concentration is easiest to see. The largest ten companies in the US market now make up around 38 to 40 per cent of the S&P 500, up from roughly 22 per cent as recently as 2020. Seven of them, often called the Magnificent Seven, account for about a third of the index on their own. Because the US is around two-thirds of the global equity market, and because most “global” funds simply follow these market weights, an ordinary diversified equity holding is, when you look through to what it actually owns, heavily invested in this small group of AI-linked giants. In a conventional balanced portfolio, once you look through to the underlying holdings, something like a fifth of the whole portfolio can sit in a dozen or so companies tied to the same theme. The “broad market” has come to behave, in large part, like a concentrated position.

The bonds you thought balanced them

The natural reassurance is that the other part of the portfolio, the bonds, provides the balance. How far that holds depends on what kind of bonds you own. The most direct overlap, where your bond fund lends to the very companies you already hold in shares, is mainly a feature of global and dollar corporate bond funds, in which the AI giants have become some of the largest borrowers as they fund the build-out with debt. Many UK investors hold exactly those, often currency hedged. It is far more muted in UK gilts, which are government debt with no direct corporate exposure, and lighter in sterling corporate bond funds, where financials and utilities dominate and the technology names are still a small presence.

What applies across all of them, gilts included, is subtler but just as important. Bonds have lately tended to fall alongside shares rather than cushion them, as gilt holders saw in 2022, because the force unsettling both has often been the path of inflation and interest rates rather than the health of the economy, and that path is now shaped in part by the build-out itself. The old reassurance, that when shares fall bonds will rise, is less dependable than it was.

The diversifiers that share the same root

Even the assets held specifically to diversify often turn out to be the same theme in a different guise. Data centres consume enormous and growing amounts of electricity, which supports investment in power and the grid. They are built from copper and other metals. They need physical space, which is part of the appeal of large logistics and warehouse property. Each of these is a reasonable long-term holding, but it would be a mistake to count them, in full, as diversification away from artificial intelligence, because in part they are exposure to it.

Adding it up

Put it together. A large share of the equity side sits in a small group of AI-linked companies. A growing share of the bond side lends to those same companies and is tied to the interest-rate path the build-out helps shape. And several of the assets held to diversify are themselves part of the theme. A portfolio that looks well spread across the standard labels, sixty per cent shares, forty per cent bonds, a little in alternatives, can, on a genuine look-through, carry a common dependence on the AI build-out far larger than those labels suggest. The diversification is real on paper. It is thinner underneath.

Seeing it is the first step

None of this is an argument against the theme. The build-out is real, the companies leading it are for the most part highly profitable and well financed, and it is worth being invested in. No investor in today’s market is free of this theme, and we would not claim to be; we believe in it, and we are invested in it. The question is not whether you hold it, but whether you hold it deliberately, and in a form you understand, rather than inheriting it in its most concentrated shape from the way the market happens to be built.

This is the kind of exposure that rarely shows up in a standard portfolio review, because a standard review looks at the labels rather than the dependencies underneath them. Seeing the true picture is the first step; shaping how that exposure is held, around each client’s own circumstances, is the work, and it is central to how we invest. If you would like us to look at your portfolio through this lens, we would be glad to talk.

This article is for information only and does not constitute personal advice or a recommendation to buy or sell any security, fund or index. The value of investments can go down as well as up, and you may get back less than you invest. Past performance is not a guide to future returns. If you are unsure as to the suitability of any investment, please seek advice that takes your personal circumstances into account.