Tech, AI and Concentration Risk in US Benchmarks
If you buy the S&P 500 in 2026, you’re not really buying “the market” anymore — you’re buying a highly concentrated bet on a handful of AI-linked mega-caps. The numbers behind that shift are stark.
Several analyses this spring estimate that AI‑related stocks now account for roughly 40–45% of the S&P 500’s total market capitalisation. A widely shared note describes how data centre, semiconductor and hyperscaler names have become the “AI backbone” of the index, with Nvidia alone at about 7% weight as of 30 March 2026, ahead of Apple at 6.3%, Microsoft at 4.6% and Amazon at 3.7%. The top five AI-centric companies now hold roughly 30% of the S&P 500, and the top 20 AI-related names nearly half — a degree of concentration that surpasses the peak of the dot-com bubble on some measures.
Goldman Sachs projects that AI infrastructure will deliver around 40% of all S&P 500 earnings growth in 2026, with capex on AI chips, data centres and related energy infrastructure on track to reach 2% of US GDP. That growth story is what supports the current valuations and index weightings. But it also creates what some strategists label the “45% fragility”: if monetisation of AI investment falls short, or if a regulatory or technological shock hits this cluster of companies, the impact on the headline index would be disproportionate.
For diversification, the implications are clear. A portfolio built around cap‑weighted S&P 500 and Nasdaq exposure in 2026 is far less diversified by underlying earnings drivers than the ticker count suggests. Sector and factor exposures are heavily skewed towards mega‑cap growth, quality and AI‑adjacent technology, with smaller weights in more traditional cyclicals, value names and domestic mid‑caps. That concentration can be a feature when AI momentum is strong, but it is a structural risk if leadership rotates or if infrastructure spending normalises more quickly than current forecasts assume.
Investors using US benchmarks as “neutral” market exposure therefore need to be explicit about what they are actually holding: in 2026, broad US indices are behaving much more like giant, liquid AI and mega‑cap funds than like equal representations of the US corporate landscape.