The Financial Times analysis finds companies supplying compute, data-center capacity and chips to OpenAI have accumulated roughly $96 billion in debt to finance rapid build-outs. The breakdown includes $30 billion already borrowed by SoftBank, Oracle and CoreWeave, with another $38 billion in talks and $28 billion in loans tied to other providers. Separately, the five major hyperscalers have taken on about $121 billion of new debt this year to scale AI infrastructure, according to Bank of America. Analysts warn the mismatch between build-out costs and near-term revenues is stressing the credit markets and raising counterparty risk. Universities that depend on cloud credits, shared compute and industry partnerships for AI research should monitor counterparties’ financial health and procurement terms: strained suppliers could alter pricing, access to GPUs, and timelines for campus research projects.