CalPERS has trimmed its exposure to the Magnificent Seven technology stocks by 26%, and its Nvidia position by nearly 40%, the pension fund disclosed. The move came as its board weighed the risks artificial intelligence poses to financial markets and to humanity more broadly, against a backdrop of soaring tech valuations. The debate at the largest U.S. public pension fund followed the resignation of a researcher at Anthropic, the AI safety company.
A stake sale timed to a valuation debate
The reductions are notable for their size. Cutting a quarter of a position in the seven biggest technology names is not a routine rebalance for a fund of CalPERS’ scale. The near-40% cut to Nvidia is sharper still. CalPERS manages retirement assets for California’s public employees, and its equity decisions are watched closely because of the fund’s size and its long horizon.
The timing matters. Tech valuations have climbed steadily on expectations that AI spending will translate into durable earnings. CalPERS’ sale lands in the middle of that run, not after a drawdown. That makes it a judgment call about risk at current prices rather than a reaction to losses.
What the board was weighing
CalPERS’ discussion covered two distinct concerns. One is financial: whether AI-related valuations have outrun the earnings that are supposed to justify them, leaving the fund exposed if sentiment turns. The other is broader: the societal and existential risks of powerful AI systems, the kind of question that until recently sat outside the scope of a pension board’s agenda.
The resignation of an Anthropic researcher gave that second concern a concrete reference point. Anthropic is one of the companies at the center of the AI buildout, so a departure there draws attention to internal disagreements about safety and direction. CalPERS didn’t tie the resignation directly to its trading decisions in the facts available.
Why a pension fund is talking about AI safety
Public pension funds hold diversified portfolios, and the Magnificent Seven have grown into a large share of broad indexes. That means a fund like CalPERS is exposed to AI-related risk whether or not it buys the stocks directly. Reducing the positions is one way to manage that exposure; debating the risk at the board level is another.
The discussion also reflects a wider shift among large asset owners. Climate risk entered pension boardrooms over the past decade. AI risk is now following a similar path, moving from specialist concern to a standing item for trustees who have to justify their decisions to beneficiaries and lawmakers.
The concentration question
The Magnificent Seven’s weight in major indexes has been a persistent worry for allocators. When a handful of stocks drive index returns, a passive portfolio becomes an active bet on those companies without the fund ever making that choice explicitly. CalPERS’ cuts suggest the fund wanted to reduce that implicit concentration.
Nvidia is the most direct AI play among the seven, which may explain the larger percentage reduction. Its chips underpin much of the data-center buildout behind current AI models. A smaller position leaves CalPERS with less upside if the buildout continues at pace, and less downside if spending slows.
What comes next
CalPERS hasn’t said whether further reductions are planned, or whether the AI risk discussion will become a formal part of its investment policy. The fund’s next disclosures will show whether the cuts were a one-time adjustment or the start of a broader shift away from concentrated tech exposure. Beneficiaries and California lawmakers will be watching those filings for signs of which way the board is leaning.
