Global pension funds cut U.S. equities on AI concentration risk
Large pension investors are reducing their exposure to U.S. equities as elevated valuations and the growing dominance of AI-linked technology stocks reshape portfolio strategy. Funds in Australia, Canada and the UK are moving below global equity benchmarks, reflecting broader concern that diversification has been undermined by market concentration.
Highlights
- Australian Retirement Trust, managing about U.S.$260 billion, reduced U.S. equity holdings in 2024, citing stretched AI sector valuations and overextended fundamentals.
- Marsh reports one-third of 430 global institutions, representing over $5 trillion in assets, plan to cut U.S. equities in the next 12 months due to tech-led concentration risk.
- People's Pension cut U.S. equity exposure from 53 per cent to 49 per cent of its main fund's global holdings, well below the MSCI ACWI index's 64 per cent weighting.
Portfolio shifts reflect valuation concerns
As reported by the Financial Times, major pension funds managing hundreds of billions of dollars are trimming U.S. equity holdings because a small group of technology and AI-related companies has come to dominate market returns.Australian Retirement Trust, which manages about U.S.$260 billion, says it has reduced its position in U.S. equities this year relative to the MSCI World benchmark. Jimmy Louca, a senior portfolio manager at the fund, says valuations in the AI sector and in U.S. equities are "a little bit stretched" and that U.S. fundamentals appear to be more than fully priced.
He says ART does not expect a dotcom-style bust, but argues that the market has moved further than fundamentals justify. That assessment is prompting the fund to stay underweight U.S. equities versus its strategic asset allocation and to seek diversification around that exposure.
Consultancy Marsh says in a report published last month that more global institutions are planning to reduce U.S. equity exposure than increase it. Of 430 entities surveyed, with more than $5 trillion in combined assets under management, one-third plan to cut U.S. equity holdings over the next 12 months, double last year's level.
Diversification gains importance across pension sector
Marsh says the heavy weighting of a handful of technology companies in major U.S. indices has made geographic exposure increasingly synonymous with sector and factor concentration. Experts argue that more than a third of the S&P 500 is now represented by large-cap companies tied to the AI investment cycle, increasing concentration risk for institutional investors.La Caisse in Canada says its U.S. allocation remains its largest, but Vincent Delisle says the fund is diversifying away from megacap technology stocks. He says valuations can become a trap and that the best risk-reward opportunities now lie outside the technology sector, while AI risk has become the single most important factor in how the pension scheme positions itself in equity markets.
In the UK, People's Pension has also cut its U.S. exposure. Public factsheets show the U.S. now accounts for 49 per cent of the global equity exposure of its main fund, down from 53 per cent at the end of last year and well below the MSCI ACWI index weighting of 64 per cent.
Dan Mikulskis, chief investment officer at People's Pension, says concentration risk merits serious discussion because the scale of the U.S. market in global portfolios and the rise of passive investing mean indices are less diversified than in the past. Denmark's ATP is also closely monitoring valuations, with chief investment officer Mikkel Svenstrup saying current prices imply very strong earnings growth expectations and leave equities more vulnerable to disappointment than under more typical market conditions.
In our earlier article on investors cutting AI concentration risk, we explained that the AI theme has become so widely embedded across U.S. stocks and even credit markets that diversification is getting harder. We noted that, in response, asset managers and clients have been increasing interest in hedge funds, private assets, emerging markets, and other lower-correlation strategies to reduce reliance on AI-linked performance.