Most people who worry about the AI boom think about jobs or electricity bills. Lenore Palladino, an associate professor of economics and public policy at the University of Massachusetts Amherst, points to a less visible exposure: pension funds and retirement accounts. In an essay for the Law and Political Economy Project, she argues that ordinary savers are financing the data-centre build-out through private funds that do not have to tell them the risks.
Her chain of reasoning runs like this. Public pension systems have moved more of their money into private credit and private equity funds, which are lightly regulated and opaque. Recent changes under the Trump administration have, she says, opened the $15 trillion in private-sector workers' retirement accounts to private funds for the first time. Those funds are where data-centre builders, and the gas plants that power them, increasingly raise money: a report from Americans for Financial Reform puts private equity behind 25-30% of data-centre financing.
The risks she cites come from credit markets rather than share prices. S&P said in September that the credit quality of the largest cloud providers is weakening, given projected spending of $7 trillion by 2030, much of it borrowed. More than $9 billion a week of credit default swaps linked to Big Tech bonds traded in September, and the protection is getting dearer. Market jitters are already visible: she notes the Magnificent Seven lost nearly a trillion dollars in value on 29 October, and the wider market lost $3.2 trillion in a single day the week after. The recent elections added a political risk, with voters backing data-centre moratoria and new governors in Texas, Georgia and New Mexico promising slowdowns.
For readers using AI tools, the connection is indirect but real. The prices of ChatGPT, Claude or Gemini subscriptions are subsidised by enormous capital spending, and if that spending is financed by debt that sours, the result could be higher prices or retrenchment. Palladino also notes that index funds are becoming more exposed as AI companies list, with Anthropic and OpenAI going public.
This is an opinion essay with a clear political view, not a financial analysis with new data. Palladino does not estimate how much retirement money is actually tied to data-centre debt, and her remedies, from trustee pressure to public asset managers, are policy proposals rather than findings. Her comparison with 2008 is a warning, not a forecast, and others argue AI demand will justify the spending. The numbers she cites come from S&P and advocacy reports, which readers can check.