Endowment and pension divestment from AI companies

Organise the withdrawal of capital by university endowments, public-sector pension funds, sovereign wealth funds, religious institutions, and mission-aligned foundations from named AI vendors (Palantir, Anduril, frontier-lab equity holdings held through public-market indices) whose products the organiser argues are structurally incompatible with the institution's stated values. The vehicle is a divestment campaign inside the institution — student coalitions at universities, member coalitions inside union pension trustees, congregational coalitions inside faith endowments — that carries the demand through the institution's own governance apparatus (trustees, investment committees, member votes) until a portfolio decision issues.

An actor chooses this strategy because it targets the institution the AI vendor's capital comes from rather than the technology's user, and it does so through a governance apparatus the vendor cannot lobby — a university's students and faculty, a union's members, a congregation's laity are not accessible to the vendor's public-affairs shop. The strategic ancestry is deep and well-tooled: the apartheid campus-divestment wave of the 1980s and the fossil-fuel divestment wave of the 2010s produced a mature organiser toolkit — resolution templates, trustee-lobbying scripts, membership-organising infrastructure — that transposes directly to AI targets. Divestment is also asymmetric to the vendor's capital: a fractional-percent equity holding withdrawn from a public-sector pension fund does not move the vendor's share price, but the organised withdrawal broadcasts a moral judgment through the institutional-investor channel that reaches other funds' investment committees and the vendor's other institutional holders.

It trades share-price-effect for narrative-and-normative effect. The withdrawal itself rarely disciplines the vendor's capital raise — a fossil-fuel-divestment scholarship shows share prices largely unaffected — and the strategy therefore lives and dies on its capacity to stigmatise the target in the eyes of the next-round investor, the next employer talent-pool, and the political coalition the vendor needs to license its next contract. It is also structurally limited to institutions that hold portfolios in the target: a movement pursuing divestment can only ever address the fraction of AI capital held by the class of institution it has organised access to, and the private-equity, sovereign-wealth, and founder-held slices route around the strategy entirely.

This strategy differs from strat-shareholder-resolutions-on-ai-vendors by direction of engagement — divestment exits the position, shareholder resolutions engage it — and from strat-institutional-procurement-refusal-of-ai-vendors by side of the transaction — divestment refuses the vendor as investee, procurement refusal refuses the vendor as supplier. Movements that run both arms hold the vendor between two market-side pressures on the same institution.