Bank of America Just Issued a Stark Warning: AI Boom Is Hitting a Cash Crunch (247wallst.com)

Bank of America warns the AI buildout is running into a cash crunch: major tech firms sharply increased borrowing for data centers in Sept–Oct, issuing roughly $75 billion in bonds and loans — more than double the decade average — while aggregate capex for AI infrastructure is projected to consume about 94% of operating cash flow (after dividends and buybacks) in 2025–26, up from 76% in 2024. Global data‑center spending estimates are staggering (Bank of America projects up to $3 trillion by 2028; McKinsey estimates $7 trillion by 2030), and companies are using a mix of corporate debt, securitizations (e.g., Blackstone’s $3.46B CMBS for QTS), private financing and off‑balance‑sheet vehicles to fund rapid expansion. Recent short‑term activity includes ~ $112B spent by Google, Meta, Microsoft and Amazon in the past three months. For the AI/ML community and investors this matters because the era of self‑funded growth is giving way to leverage-driven expansion, increasing financial risk if AI returns or adoption slow. Company-level differences are material: Nvidia’s low debt and huge free cash flow make it resilient, Meta’s large cash buffer helps absorb new financing, while Oracle’s and smaller players’ heavy borrowing could push interest expenses into earnings and compress ROI on capex. Practical takeaways: monitor interest‑coverage ratios, capex ROI and financing terms; accelerated deployment can speed model scale and research, but systemic leverage raises tail‑risk for valuations and long‑term infrastructure economics.