AI capex is surging — is my portfolio too exposed to the AI builders?
Hyperscaler AI capex has become one of 2026's defining market debates — and whether it's justified or overbuilt matters less to your portfolio than a more basic question: how much of it already depends on a handful of the same companies. Global index funds are more concentrated in AI-driven mega-caps than they've been in years, and investors who also hold individual positions in those names can end up far more exposed than they think.
The debate dominating 2026
Hyperscaler capital expenditure — the money Microsoft, Google, Amazon, and Meta are spending on data centres, chips, and power to build AI infrastructure — has become one of the defining questions of the market cycle. Combined capex from the largest AI builders is running into the hundreds of billions of dollars a year, funded partly by cash flow and partly by a growing web of vendor financing and circular deals between chipmakers, cloud providers, and AI labs. The bull case is that this spend is building the infrastructure layer for the next decade of productivity growth. The bear case is that returns on this capital are unproven, the financing structures are opaque, and a slowdown in AI revenue growth would leave a lot of expensive, depreciating hardware without the earnings to justify it. Nobody — including the companies spending the money — knows for certain which case is right.
Why this matters even if you've never bought a single AI stock
Most UK DIY investors don't hold individual AI stocks — they hold global index trackers, and assume that spreads their risk. It doesn't spread it as much as it used to. Market-cap weighting means an index fund holds more of whatever has gone up the most, and the handful of companies driving AI capex — the same names at the centre of the debate above — now make up a historically large share of funds like the MSCI World and S&P 500. A "globally diversified" tracker today has a meaningfully larger single-theme concentration than the same tracker had a decade ago, simply because those companies have grown so much faster than everything else in the index.
Owning the trade via an index vs being unknowingly overconcentrated
These are genuinely different things, and the difference is intent. Choosing to hold a market-cap-weighted global tracker and accepting that it now leans heavily on a handful of AI-driven mega-caps is a legitimate, informed decision — that's how index investing has always worked, and trying to strip out whichever sector is currently leading is its own form of market timing. The problem is when investors don't realise how concentrated their "diversified" fund already is, and then add individual positions in the same handful of companies on top — a popular AI stock bought directly, sitting alongside a global tracker that already holds a large weighting in that same stock. The combined, real exposure can be far higher than either holding looks on its own.
A simple framework: trim, hold, or diversify
- Hold if your only AI/mega-cap tech exposure comes through a single broad, low-cost global or US index fund, and you're comfortable that market-cap weighting means owning more of whatever is currently leading. This is the default, and there's nothing wrong with it.
- Trim if you hold individual AI/tech stocks on top of an index fund that already has a heavy weighting in the same names — check your combined single-stock exposure, not just what each holding looks like in isolation.
- Diversify if you want less dependence on a handful of mega-cap outcomes without trying to predict when or whether AI capex disappoints — adding funds with different weighting methods (equal-weight, value-tilted) or more exposure to markets outside the US spreads the same total equity allocation across more independent bets.
Key takeaway: The AI capex debate will resolve one way or another, but you don't need to predict the outcome to check something more basic first: how much of your portfolio's fate is already tied to a handful of companies, and whether you chose that deliberately or arrived at it by accident through index weighting and individual holdings stacking on top of each other.
Arken maps your true exposure across every fund and stock you hold — including what your index trackers already hold on your behalf — so a concentration in AI-driven mega-caps shows up clearly, whether it came from one holding or several stacked on top of each other.
