Capital Group (July 23, 2026) argues that AI-driven market concentration has reached levels not seen in decades, creating unintended risks in cap-weighted index funds. The piece uses four charts and commentary from CIO Martin Romo and portfolio managers Brady Enright, Steve Watson, Damien McCann, and Jody Jonsson.
Key takeaways
- Markets are among the most concentrated in recent history.
- AI concentration is a global phenomenon.
- U.S. GDP relies heavily on AI spending.
- AI-related issuance is straining parts of the bond market.
1. Extreme U.S. concentration, but a single theme
Top 10 stocks are approaching 40% of the S&P 500—similar to 1964’s Nifty Fifty peak (39%) and Japan’s 1980s boom. Then the leaders spanned sectors (AT&T, GM, Exxon, IBM). Today NVIDIA, Microsoft, Amazon, Micron and peers are tightly linked to one theme: AI capex. That raises correlated-downside risk (already visible in early-July pullbacks in SK hynix and Sandisk). Past concentration episodes eventually unwound; the authors treat this as a reminder, not a crash forecast. Quality names left behind (Royal Caribbean, P&G, Citigroup) trade at 20%+ discounts to history.
2. Same pattern globally
MSCI Emerging Markets top 10 = 41% of the index; SK hynix + Samsung + TSMC alone = 29%. Developed non-U.S. and Europe are less extreme. Watson sees value in non-U.S. leaders (AstraZeneca, Tencent) and “AI wreckage” names he views as enablers rather than victims (SAP, Trip.com, Amadeus).
3. Economy-level dependence
St. Louis Fed estimates: AI-related investment added nearly 1% to U.S. real GDP in the first nine months of 2025—39% of total growth. Roughly half of data-center costs are semiconductors, but HVAC, power, water and transformers also benefit. Enright’s next focus is companies using AI for durable advantage (financials, healthcare) rather than just selling the build-out.
4. Bond-market strain
Hyperscalers (Alphabet, Amazon, Meta, Microsoft, Oracle, SpaceX) now represent ~4.8% of the Bloomberg U.S. IG Corporate Index—up 78% year-over-year. Heavy issuance has pressured some of those credits. McCann stays constructive on broader credit (earnings, consumer, labor) but stresses issuer-by-issuer selection and diversification across IG, HY, securitized and EM.Bottom line
Index funds are cheaper, not safer. Passive portfolios now embed a large bet that one AI outcome dominates. Authors recommend checking how much of a portfolio sits in a handful of AI-linked names, adding old-economy and non-U.S. exposure, and being willing to differ from the benchmark. “Bold enough to own great companies when fundamenta
Leave a Reply
You must be logged in to post a comment.