Nearly six in 10 companies on the S&P 500 rose in July, yet the index still recorded its first negative July since 2014. That disconnect highlights what the Bank of England sees as a growing risk: market performance is becoming increasingly dependent on a small number of AI-related companies.

In its July 2026 Financial Stability Report, the BoE noted that the S&P 500 now accounts for around half of global equity market capitalisation. More strikingly, AI-related companies now represent roughly half of the index's value, compared with around a quarter in 2022. The concern is not that AI is a bubble, it is that an increasing share of global equity returns is tied to a single theme.

If expectations for AI earnings were to weaken, the impact could be amplified by crowded positioning and leverage. But even if AI continues to exceed expectations, investors still face the risks that come with growing concentration. Put simply, a larger share of market performance is becoming dependent on a handful of companies. And many portfolios are less diversified than investors think.

Investors can't ignore concentration risk any longer

Concentration risk – the danger that there is too much weight in a small number of investments, companies or themes – is no longer a theoretical concern. It is already shaping market behaviour.

Technology shares have been repeatedly volatile as investors have reassessed AI spending, infrastructure investment and future earnings expectations, and July provided one of the clearest examples yet.

The index recorded its first negative July in 12 years, with weakness concentrated among the largest technology and semi-conductor names. The iShares Semiconductor ETF fell 22.1%, its worst month since 2002. Meanwhile, the average company generated positive returns even as the index declined. The message was clear: earnings were not the problem – concentration was.

The dynamic worked both ways. Just as weakness in a handful of AI-related companies dragged markets lower, strong results from Microsoft helped drive a sharp rebound across major US indices at month-end. When the earnings announcement of a single company can materially influence the direction of the wider market, concentration risk becomes difficult to ignore.

AI concentration risk in UK investments

It is easy to see this as a US issue, but that would be a mistake.

Millions of UK savers are invested in the same companies through pensions, ISAs, model portfolios and global trackers. As portfolios become more global, exposure naturally shifts towards the largest US technology firms.

Many investors therefore believe the risk sits in Silicon Valley, when, in reality, it already sits within their pension, ISA or investment portfolio.

Why diversified portfolios may not be diversified

Research from firms including JP Morgan Asset Management and RBC Wealth Management highlights the scale of the shift. Depending on the methodology used, the largest seven to 10 companies now account for between a third and more than 40% of the S&P 500.

Market-cap weighting means investors can look diversified while remaining exposed to the same underlying drivers. That challenges one of the investment industry's core assumptions about the reduced exposure of diversified portfolios, showing that diversification on paper does not always mean diversification in practice.

Hidden AI exposure could be spreading across portfolios

The concentration story does not end with listed equities.

The AI boom is increasingly being financed through private credit, high-yield debt and infrastructure investment. What drives valuations in public markets is also driving demand for the financing of data centres, energy infrastructure, cloud capacity and AI development.

As a result, investors may be exposed to the same underlying growth story through multiple parts of a portfolio. On paper, a portfolio may appear to be diversified across asset classes but, in reality, it is concentrated around AI, and its performance may depend on the continued growth and profitability of the same AI ecosystem.

Diversification only works when exposures respond differently to changing market conditions. If enthusiasm for AI weakens, seemingly different parts of a portfolio may come under pressure at the same time because they are linked to the same underlying driver.

How wealth managers can assess AI concentration risk

The real question is not whether AI succeeds or fails, it is whether firms understand how much of a client's future returns depend on it.

This is where the discussion moves beyond market commentary and into governance, suitability and consumer duty. Firms should understand the true concentrations embedded within portfolios. They should undertake look-through analyses and ensure client communications accurately reflect the exposures being taken.

Understanding concentration may soon matter as much as selecting investments.

The bigger question: what does true diversification look like?

The Bank of England is not questioning the future of AI, it is questioning the resilience of markets that have become increasingly dependent on it.

For wealth managers, the key issue is not valuation, it is concentration.

Many investors believe they are diversified because they own multiple funds, strategies and asset classes. But the greatest portfolio risk is not the one that investors can see, it's the one they think they have already diversified away.

For more on diversification of portfolios and market insight for wealth managers, please contact Amy Shooter.

Latest related insights