Project Finance versus Corporate Finance
There are two fundamentally different ways to borrow against a data center, and the distinction shapes everything downstream. The first is corporate finance: a company borrows on the strength of its whole balance sheet. A large operator issues a bond or draws a credit facility backed by all of its assets and cash flows, and lenders look to the enterprise as a whole. If one building underperforms, the rest of the company still stands behind the debt. This is how investment-grade operators and the hyperscalers themselves fund much of their activity. It is fast, flexible, and cheap for a strong credit, because the lender is buying the diversified risk of an entire firm.
The second is project finance, and it is the workhorse of large infrastructure. Here the developer creates a separate legal entity, a special purpose vehicle, that owns exactly one thing: this campus. The debt is borrowed by that vehicle and secured only by that campus and its lease. The loan is non-recourse or limited-recourse, meaning that if the project fails, the lenders can seize the asset but generally cannot pursue the parent company's other assets. Risk is ring-fenced inside the box. The lender underwrites the project on its own merits: the tenant, the lease, the power, the construction plan, and nothing else.
The trade-offs are sharp and worth memorizing. Corporate finance is simpler, faster to execute, and usually cheaper for a strong balance sheet, but it consumes the borrower's overall debt capacity and puts the parent on the hook. Project finance is slower and more expensive to arrange, loaded with legal and advisory cost, but it isolates risk, keeps the debt off the parent's primary balance sheet, and lets a mid-sized developer punch far above its weight. A developer worth a few hundred million can build a billion-dollar campus because the project stands on its own tenant and power contracts, not on the developer's net worth.
Project finance also disciplines the deal. Because repayment depends entirely on this one asset's cash flow, lenders impose tight structure: cash flows run through controlled accounts, reserves are funded for maintenance and debt service, and distributions to the equity owner are blocked until obligations are met. The vehicle is engineered so that money hits the lenders before it reaches the sponsor. That waterfall is the reason non-recourse debt can be priced reasonably despite having no parent guarantee behind it.
The choice between the two is driven by who the sponsor is and what they want. Hyperscalers with fortress balance sheets often prefer corporate finance or even self-funding for speed and control, and increasingly they keep the most strategic capacity on their own books. Independent developers and infrastructure funds lean on project finance because it lets them recycle a small equity base across many deals. A common hybrid has emerged: the developer builds and finances a campus in a project vehicle, signs a long lease to a hyperscaler, and the lease's credit quality does the heavy lifting even though the tenant never guarantees the debt directly.
A third pattern sits between the two: the joint venture. A deep-pocketed capital partner, often an infrastructure fund or pension, supplies most of the equity alongside the developer, and the vehicle then raises project debt on top. This spreads the equity risk, brings in a partner with a low return hurdle, and still isolates the asset. The result is a structure where cheap institutional equity and cheap project debt combine to fund an asset that neither party would build alone.
The practical lesson is to read a financing announcement structurally. When a developer says it has secured funding for a campus, ask whether the debt is recourse to a parent or ring-fenced in a project vehicle, and who supplied the equity. That single distinction tells you who actually bears the risk, how much more the sponsor can build, and how fragile or durable the structure is if the asset stumbles. Structure is destiny.