AI Borrowing Creates a New Credit Playbook

The AI-driven capital expenditure cycle is expanding at an unprecedented scale and speed, with projected spending for the five largest hyperscalers revised upward from $450 billion to $1.2 trillion by 2027. This surge is fueled by a 350% increase in global weekly token usage since early January, signaling massive demand for compute power. Credit markets are rapidly evolving to finance this ecosystem, moving beyond traditional US dollar investment-grade bonds into diverse currencies and innovative structures like GPU-backed asset-based financing and high-yield issuance for "Neo clouds." However, physical and regulatory bottlenecks—specifically power grid access, equipment shortages, and permitting delays—are emerging as critical gating factors. As electricity availability becomes a primary constraint, energy infrastructure financing is increasingly merging with AI infrastructure credit. This structural shift marks a defining era for credit markets, characterized by a blurring of boundaries between public and private capital.