BIS General Manager Pablo Hernández de Cos said capital spending by the largest technology companies was outpacing their cash flows, pushing AI investment towards debt and private credit. The five biggest companies are expected to spend more than 1 trillion dollars between 2025 and 2026. Opaque and interconnected financing could amplify a correction if returns fail to meet expectations.
Economics & Markets··Evening
Spending has overtaken cash flow
Pablo Hernández de Cos, general manager of the Bank for International Settlements, the institution often described as the central bank for central banks, set out the financing structure of the AI investment boom at the Global Fintech Fest in Mumbai. He said capital spending by the largest technology companies was outpacing their cash flows. The BIS speech and Reuters reporting published by MarketScreener both said funding was shifting from corporate earnings towards debt and increasingly private credit.[1], [2]
Five companies are set to spend over 1 trillion dollars
The BIS estimates that the five largest technology companies will allocate more than 1 trillion dollars to AI-related capital expenditure between 2025 and 2026. Hernández de Cos said industry participants expected global investment to rise from about 500 billion dollars today to between 3 trillion and 4 trillion dollars by 2030. Reuters also reported the same figure of more than 1 trillion dollars for the five largest companies.[1], [2]
Circular financing obscures the links
Hernández de Cos described much of the financing as opaque and interconnected. Under circular financing, chipmakers and large cloud providers take equity stakes in AI companies, which in turn commit to buying their investors' chips and computing capacity. He said investment could retreat if returns disappointed, while a sharp equity repricing could pass more forcefully into household consumption. Reuters reported his warning that these financing links merited close scrutiny.[1], [2]
Related columns
For more information on this topic, you can read the related columns.
