Oracle secures $38B debt deal to finance two new data centers in Texas and Wisconsin
Oct 24, 2025
Key Points
- Oracle secures $38 billion in debt to build data centers in Texas and Wisconsin, marking the first major hyperscaler to finance AI infrastructure through debt rather than cash flow.
- Meta, xAI, and other players are following suit as debt markets warm, with favorable borrowing rates creating room for massive capital deployment.
- The shift to debt financing introduces structural risk: hyperscalers now face material ROI consequences, and failure to monetize assets could expose retail retirement accounts to losses.
Summary
Oracle secured a $38 billion debt deal to finance two new data centers, allocating $23 billion to Texas and nearly $15 billion to Wisconsin. The move marks a shift in how hyperscalers are funding AI infrastructure buildout.
Most AI capital spending has been financed from cash flow rather than debt. This deal represents a material inflection point. An observer notes that a bubble typically emerges once the majority of expansion shifts to debt financing, suggesting Oracle's move may signal that threshold approaching.
The debt financing landscape for AI infrastructure is already warming. Meta has tapped debt markets, xAI has used debt, and multiple players are following suit. Microsoft's 10-year bonds trade at only 3 basis points above the risk-free rate, giving hyperscalers substantial room to issue debt or access private credit.
This shift introduces structural fragility. Additional leverage unlocks more spending, but it also means consequences for failing to achieve ROI become material in ways they have not when funding was self-generated. The calculus changes when a school teacher's 401(k) holds the debt. One junk bond investor noted the darker implication of offloading data center construction debt into retail retirement accounts.
Monitoring the share of AI CapEx financed through debt versus cash flow is now the most important metric for assessing where the cycle actually stands.