The Bank for International Settlements published a BIS Bulletin examining how the artificial intelligence investment boom is affecting macroeconomic outcomes and corporate financing, concluding that the scale of anticipated spending is pushing firms to move from funding capital expenditure with operating cash flows to greater use of debt, including private credit. It assesses macroeconomic and financial stability risks as moderate for now, but argues that the boom’s sustainability depends on AI firms meeting high earnings expectations, with equity prices having run well ahead of debt-market pricing. In the United States, AI-related investment in semiconductor manufacturing facilities and data centres has risen from a negligible contribution before 2022 to adding about 0.4 percentage points to GDP growth on average over the subsequent three years, while total information technology investment has accounted for almost half of GDP growth in recent quarters. By mid-2025, spending on IT manufacturing facilities and data centres, including construction and equipment, was equivalent to 1% of GDP and total IT-related investment, including other IT equipment and software, reached 5% of GDP, exceeding its peak during the dot-com boom and driven mainly by IT-producing firms. Analyst forecasts cited in the Bulletin indicate annual data centre spending could increase by between USD 100 billion and USD 225 billion in the next five years, taking it to 0.8% to 1.3% of GDP from about 0.5% currently. The Bulletin notes that leading technology firms driving AI investment, including Alphabet, Amazon, Meta, Microsoft and Oracle, have historically operated with less debt than other firms but have ramped up capital expenditures, with free cash flows recently lagging capex, increasing the need for external funding. Financing is increasingly provided through debt instruments such as corporate bonds, leasing arrangements and loans, while project risks related to construction, power availability and tenant concentration can place some deals outside traditional bank and bond markets. Private credit outstanding to AI-related companies has grown from near zero to over USD 200 billion and to almost 8% of private credit loan volumes, and the Bulletin estimates USD 300 billion to USD 600 billion outstanding by 2030 based on projected AI-investment growth. Loans to AI-related borrowers are similar to other private credit loans in secured share at 46% versus 48%, maturity at 4.7 versus 4.8 years and spreads at 6.2 versus 6.1 percentage points over the London Interbank Offered Rate or the Secured Overnight Financing Rate, but are larger on average at USD 169 million versus USD 90 million, and only about 20% of private credit funds have AI exposure with an average allocation of around 5% of lending volumes.

Original source View in tracker