Global Pension Funds Trim U.S. Stock Holdings as AI Concentration Risk Becomes Core Concern
nashnova research
Several global pension funds managing a combined trillion-plus dollars have actively cut U.S. stock holdings below benchmark — the core driver is AI-related stocks dominating U.S. indices, paired with stretched valuations. This means the world's most patient long-term capital is voting against the AI premium in American equities.
Who is selling, and how much?
Australia's AustralianSuper (roughly $260 billion under management) has trimmed U.S. equity below its MSCI World benchmark this year.
Canada's CPP Investments (roughly $388 billion) still holds the U.S. as its largest single-market exposure but is actively diversifying away from mega-cap tech.
The UK's People's Partnership (roughly £45 billion) cut U.S. stocks from 53% to 49% of its global equity allocation — well below MSCI World's 64% U.S. weight.
This means → three top-tier pension funds in three different countries made the same move in the same window: reduce U.S. equities, reduce tech concentration.
What exactly are they worried about?
The central concern is AI concentration risk: more than one-third of the S&P 500's weight sits in large-cap companies tied to the AI investment cycle. In plain terms = when you think you're buying "the U.S. market," you're actually betting on a handful of tech firms' AI narrative.
Consultancy Marsh surveyed 430 institutions (managing over $5 trillion combined); one-third plan to cut U.S. equity exposure in the next 12 months — double the share from a year ago.
Marsh's report states: "The high concentration of a few technology names in major U.S. indices means geographic exposure has become almost synonymous with sector and factor concentration."
Are valuations actually stretched?
AustralianSuper portfolio manager Jimmy Louca said U.S. fundamentals are "fully — if not excessively — priced." Market gains have exceeded what fundamentals can support.
Denmark's ATP chief investment officer Mikkel Svenstrup noted current valuations embed exceptionally strong earnings-growth expectations for years ahead. This means → the stock price already bakes in the most optimistic script; any earnings miss would trigger a steeper fall than in a normal market.
Yet Louca also said he does not expect an "internet-bubble-style crash." This reflects a nuanced institutional stance: "expensive, diversify" — not "collapse, liquidate."
Has diversification hurt performance so far?
CPP Investments head of investments Vincent Delisle acknowledged that diversification has been "punished" because tech stocks carry heavy index weight and have kept outperforming.
In plain terms = over the past few years, anyone who trimmed U.S. tech underperformed the benchmark — so these funds are making this call against short-term performance pressure.
Delisle nonetheless argues "diversification and patience remain the right choice" — his reasoning: valuations may now be a trap, and the best risk-reward opportunities lie outside the tech sector.
What does this mean for ordinary investors?
People's Partnership CIO Dan Mikulskis warned: "Indices are not as diversified as they once were." This means → passive investors — people who simply buy index funds — may be carrying far more concentration risk than they realize.
These pension funds manage retirement savings over horizons of decades. Their collective action signals that long-horizon institutional skepticism about AI valuations has shifted from discussion to execution.
The core logic: whether U.S. equity valuations hold up depends on whether AI-linked companies can deliver the high earnings already priced in — if they can't, today's concentration becomes an amplifier on the downside.
市场有风险，内容仅供研究参考，不构成投资建议。