The $5 Trillion Build-Out: Debt Markets Shoulder the AI Burden
The financing of the Artificial Intelligence revolution has shifted phases. While 2024 was defined by massive equity valuations for chipmakers, the narrative for late 2025 and early 2026 has turned decidedly toward debt. The physical infrastructure required to power the AI boom (data centers, power grids, and cooling systems) is generating a tsunami of borrowing demand. Recent reports indicate that the credit market has been hit with a $200 billion flood of AI-related issuance, signaling a massive capital rotation from software speculation to hard asset construction.
Major financial institutions, including JPMorgan Chase, now estimate that the total bill for this infrastructure build-out could reach between $5 trillion and $7 trillion over the next five years. As equity markets price in perfection, the credit markets are being asked to finance the steel, concrete, and energy that will make these digital ambitions a reality.
The “Off-Balance Sheet” Maneuver
A distinct trend has emerged among the technology giants: risk shifting. Companies like Microsoft, Meta, and Google are increasingly utilizing complex financing structures, such as synthetic leases and partnerships with private developers, to keep massive capital expenditures off their corporate balance sheets. According to the nytimes.com, these deals allow tech titans to secure computing power without directly holding the debt, effectively pushing financial exposure onto smaller developers and private lenders.
This structure creates an opaque layer of risk within the financial system. While the “Hyperscalers” maintain pristine credit ratings, the entities actually building the facilities are often privately held and less transparent. fortune.com reports that some borrowers are requesting loan-to-value ratios as high as 150%, justified only by aggressive projections of future rent growth. This exuberance mirrors the telecom fiber boom of the late 1990s, a period characterized by overbuilding and eventual consolidation.
Private Credit Fills the Gap
With traditional banks constrained by regulatory caps and exposure limits, private credit has stepped in as the primary engine for this expansion. Data from bnnbloomberg.ca suggests that private credit AI loans nearly doubled over the last 12 months. Morgan Stanley estimates that private markets could supply over half of the $1.5 trillion specifically needed for data center construction through 2028.
This shift validates the private credit thesis: flexible, non-bank capital is essential for complex infrastructure projects that do not fit the rigid underwriting boxes of traditional chartered banks. However, it also demands rigorous discipline. As capital floods into the sector, the spread between high-quality infrastructure investments and speculative projects is narrowing dangerously.
The Bond Capital View: Disciplined Allocation
The current market state is undeniably “Risk-On,” fueled by the belief that demand for compute power is insatiable. While Bond Capital acknowledges the generational shift AI represents, we remain cautious of the “irrational exuberance” described by market veterans like Howard Marks. The physical build-out of AI requires massive capital, but not all square footage is created equal.
We focus on the underlying collateral (the real estate and the power contracts) rather than the speculative value of the algorithms running inside. While banks aggressively loosen terms to win volume in this “gold rush,” Bond Capital maintains strict credit discipline. We prioritize lending against tangible assets with clear covenants, ensuring that our investors are protected even if the projected demand for AI capacity normalizes. In a market priced for perfection, certainty of execution and asset protection remain our paramount objectives.
