UAO Research · May 2026 · AI & the Long-Term Portfolio · A Universal Asset Owners special report
Artificial intelligence has stopped being a thematic sleeve and become a structural feature of the index. A handful of AI-exposed megacaps — the "Magnificent Seven" — now make up roughly 35% of the S&P 500, up from around 12% a decade ago, and the top ten companies account for more than a third of a $50-trillion-plus index (The Magnificent Seven's market cap vs. the S&P 500, The Motley Fool.). For a universal owner that holds the market by construction, this is not a stock-picking question. It is the uncomfortable realisation that a single technology bet has been embedded into the core of the passive portfolio, whether the fund chose it or not.
What the evidence says
The exposure shows up in three distinct ways. The first is concentration. A market-cap-weighted global equity portfolio is now more concentrated in a few names than at any point in modern index history; an allocator who believes they are diversified across thousands of companies is, in practice, making an outsized bet on the earnings of a small cluster of AI-leveraged firms. The second is the capital-expenditure supercycle. The largest hyperscalers committed over $300 billion to AI infrastructure in 2025, and the five biggest US cloud and AI providers have signalled $660–690 billion of capex for 2026 — roughly double the prior year (Why AI companies may invest more than $500 billion in 2026, Goldman Sachs.). That spending is simultaneously a demand shock for chips, power and data centres, and a test of the return on invested capital underneath the whole trade. The third is the productivity question: whether AI delivers an economy-wide step-change in output — which would lift the entire portfolio — or a narrower, slower diffusion that mostly redistributes profit among incumbents.
The honest position is that these three forces point in different directions. Concentration is a risk; the capex cycle could be either an investment in future cash flows or a value-destroying arms race; and the macro upside is real but unevenly timed.
Where it is contested
This is the most genuinely disputed terrain in markets. Optimists argue that AI is a general-purpose technology whose productivity gains will compound for decades and justify the spending. Skeptics — including prominent academic economists — argue that the near-term productivity uplift will be modest and that current capex implies revenue that may not arrive on schedule, raising the specter of a classic overbuild. There is no settled number for AI's contribution to total-factor productivity, which is precisely why reasonable institutions hold opposite views. And there is a reflexive risk a universal owner cannot ignore: because the AI leaders are now such a large share of the index, a sharp repricing of the AI trade would be a repricing of the whole equity book, not a sector rotation.
From the allocator's seat
The universal-owner response is not to forecast the winner but to manage the exposure deliberately. That starts with measuring true concentration — look through the index to see how much of the portfolio's risk now sits in a handful of correlated names, and decide consciously whether to accept, cap or hedge it (equal-weight, capped or factor-tilted sleeves are the usual tools). It extends to the supply chain: the most durable AI exposure for a long-horizon owner may be in the picks-and-shovels — power generation, grid, data-centre real estate and the physical infrastructure the capex cycle requires — which a universal owner can access through private and real-asset programs. And it argues for stress-testing the portfolio against an AI-capex disappointment, not just an AI boom, because the asymmetry now runs through the core holding.
What to watch next
Watch hyperscaler capital-expenditure guidance through 2026 and, more importantly, the first signs of capex discipline or a return-on-investment reckoning. Watch index concentration metrics — the share of the S&P 500 and MSCI ACWI in the top ten — for whether the trend extends or reverses. Watch power: data-centre electricity demand is the clearest physical proxy for whether the build-out is real. And watch how the largest sovereign and pension funds disclose AI exposure, the leading edge of treating it as a portfolio-level risk rather than a sector view.
Sources
- The Magnificent Seven's Market Cap vs. the S&P 500, The Motley Fool.
- Why AI Companies May Invest More than $500 Billion in 2026, Goldman Sachs.
- Big Tech AI Spending Over Time (2022–2025), Visual Capitalist.
- Executive summary — Energy and AI, International Energy Agency.
UAO Research. AI-assisted monitoring and drafting; reviewed and edited by the UAO editorial desk before publication. Not investment advice.