AI Boom Fuels Risky Bet: Data Centers Built on GPU Collateral

Banks and investors are increasingly financing AI data centers by using graphics processing units (GPUs) as collateral, according to a new report. This approach marks a shift from traditional lending practices. GPUs are the powerful computer chips that train and run artificial intelligence models. They are expensive and depreciate quickly as newer models are released. Under this novel financing method, lenders accept GPUs as security for loans. If a borrower defaults, the lender can seize and sell the chips to recover losses. The practice raises concerns. GPU values can drop sharply when next-generation chips arrive. This could leave lenders with assets worth far less than expected. Despite the risks, demand for AI computing power continues to surge. This drives companies to seek creative ways to fund costly data center construction. Industry analysts say the trend reflects both the enormous opportunity and the uncertain risks in the rapidly growing AI sector.