The Shift from Code to Concrete
The artificial intelligence narrative has undergone a structural shift as 2025 draws to a close. While early headlines focused on generative models and equity valuations, the story has moved decisively to the balance sheet. The physical build-out of the AI ecosystem—the steel, concrete, and copper required for data centers—is now a credit story of historic proportions.
According to recent data, AI project financing deals have surged to $125 billion year-to-date, a nearly tenfold increase from the $15 billion recorded during the same period in 2024. A report by [investmentnews.com](https://www.investmentnews.com/equities/ais-5t-data-center-boom-will-dip-into-every-debt-market-jpmorgan-says/262970) highlights that AI-linked companies now account for 14% of the Investment Grade Index, holding $1.2 trillion in debt and surpassing U.S. banks as the dominant sector.
This massive rotation of capital signifies that the industry is moving from an R&D phase into an industrial deployment phase. However, as capital expenditure requirements balloon, public markets are showing signs of saturation, forcing borrowers to seek liquidity beyond traditional banking channels.
Yellow Flags in Public Markets
Despite the boom, volatility is beginning to surface in public credit markets, suggesting that the initial “risk-on” euphoria is tempering. Notable cracks have appeared in the credit profiles of major technology stalwarts. For instance, credit default swaps (CDS) for Oracle—effectively insurance against default—recently spiked to their highest levels since 2009. As reported by [bnnbloomberg.ca](https://www.bnnbloomberg.ca/business/artificial-intelligence/2025/12/12/five-debt-hotspots-in-the-ai-data-centre-boom/), this pricing adjustment reflects growing investor concern over the sheer scale of debt required to fund these infrastructure ambitions.
Further complicating the landscape are “circular deals” and off-balance-sheet financing structures that have drawn scrutiny from regulators, including the Bank of England. With borrowers reportedly seeking loans covering up to 150% of construction costs, fears of a supply glut are growing. Oaktree Capital Management’s Howard Marks noted in a recent memo cited by [fortune.com](https://fortune.com/2025/12/12/ai-data-center-boom-sparks-fears-of-glut-amid-lending-frenzy/) that disciplined lending is becoming scarce, raising the risk that today’s boom could lead to tomorrow’s distressed assets.
The Private Credit Imperative
As public markets grapple with valuation concerns and regulatory scrutiny, the financing gap is widening. Morgan Stanley estimates that private credit markets could supply over half of the $1.5 trillion needed for the data center build-out through 2028. This environment favors lenders who can underwrite complex, asset-heavy projects rather than relying solely on corporate cash flows or financial engineering.
What This Means for Borrowers
For mid-market operators and developers, the implications are two-fold:
- Capital Availability: While liquidity is abundant, the cost of capital is bifurcating. Prime assets with clear revenue visibility will attract aggressive terms, while speculative projects face a steep risk premium.
- Focus on Fundamentals: The market is moving away from funding “concepts” to funding “capacity.” Lenders are prioritizing tangible collateral—land, power rights, and hardware—over pro-forma growth projections.
At Bond Capital, we recognize that the AI infrastructure boom is a physical reality that requires flexible, disciplined capital. While public markets oscillate between fear and greed, our focus remains on the underlying asset value and the certainty of execution. In a market priced for perfection, credit discipline is the ultimate hedge.
