AI infrastructure spending remains a leading perceived credit risk as Apollo Global Management points to a roughly 60-basis-point gap between credit default swaps on hyperscaler debt and bank debt since October 2025. Chief economist Torsten Slok said the widening reflects weakening credit fundamentals rather than dealer hedging, citing debt-financed AI capital expenditures, rising leverage, negative free cash flow and uncertain returns from depreciating assets. Alphabet and Meta each have forward negative free cash flow near $25 billion, while Amazon's is near $30 billion; Microsoft remains cash-flow positive, FactSet data cited in Slok's note showed. Bank of America's September survey had already identified AI infrastructure spending as fund managers' leading potential source of a systemic credit event. The outlook depends on whether AI-related revenue catches up with the investment cycle, particularly after leading AI labs called for slower model development over safety concerns.