### The AI Infrastructure Debt Boom
The race to build the physical backbone of artificial intelligence is triggering a historic surge in corporate borrowing. In 2025 alone, borrowing for data center construction skyrocketed 112% year-over-year, reaching $25 billion. This frenetic pace is driven by the specialized demands of AI workloads: massive facilities requiring advanced cooling, redundant power, and racks of high-performance GPUs that depreciate faster than any traditional industrial asset.
According to new data from **JPMorgan**, the total bill for the AI data center build-out could hit $5 trillion over the next five years. While hyperscalers like Meta and Oracle are leading the charge, issuing billions in investment-grade bonds, a significant funding gap remains. Analysts at **Morgan Stanley** project a capital expenditure of $2.9 trillion through 2028, with roughly half requiring external financing beyond company cash flows.
### The Shift to Flexible Capital
Traditional equity markets cannot single-handedly fund this expansion. As a result, the industry is turning to the debt markets. The boom is increasingly filling the order books of **private credit funds** and **asset-backed security (ABS)** issuers.
Key drivers of this trend include:
* **Liquidating Cash Flows:** By 2026, AI capital expenditures are expected to consume up to 94% of operating cash flow for major tech players, forcing reliance on debt.
* **Asset-Based Finance (ABF):** Lenders are structuring deals based on the physical value of the data centers and the leases within them, though this comes with risks.
* **Hardware Obsolescence:** Unlike traditional real estate, the core asset here (GPUs) effectively loses its value in 18 months, creating a unique “duration mismatch” for lenders accustomed to 30-year depreciation cycles.
### The Risks of Speculative Building
A “borrow-to-build” race has emerged, with developers breaking ground on facilities before securing tenants, hoping to capitalize on future demand. This speculative strategy exposes the market to refinancing risks if interest rates rise or if AI monetization delays create cash flow gaps.
***
### Bond Capital Take
“The physical build-out of AI requires massive capital. While equity grabs headlines, debt finances the steel and concrete. We understand complex capex needs. As the asset class scales, manager selection becomes critical; generalists struggle to price the risk of hardware obsolescence, while specialists outperform by structuring deals that account for the unique lifecycle of AI infrastructure.”
