An FT analysis finds companies supplying compute, data‑center capacity and chips to OpenAI have taken on about $96 billion of debt to scale operations, a liability that raises new risks for universities that partner on compute‑intensive research or rely on third‑party infrastructure. The debt load—borne by hyperscalers and specialist providers—could increase costs for academic cloud and AI collaborations and influence grant budgeting for compute‑heavy projects. Administrators and research offices must now weigh counterparty and infrastructure risks when negotiating access to AI compute resources for campus labs and centers.