How Data Center Financing Rewrites The Rulebook

Surging AI demand is creating unprecedented demand for capital. Here we explore how data center financing structures are evolving to address the different stages involved in infrastructure development, from land acquisition and construction through to long-term capital markets’ takeout.

In brief

• Surging AI demand is driving a major need for new data center capacity, creating significant opportunities for banks, institutional investors and private credit providers to fund digital infrastructure.
• Financing structures are becoming more sophisticated as projects move from land acquisition and construction through to operational, income-generating assets, requiring a blend of real estate, infrastructure and structured credit approaches.
• AI-focused data centers add complexity because developers may need to finance both the physical “shell” and high-value “core” equipment, with hardware obsolescence and shorter equipment lifecycles creating additional risk.
• Portfolio financings can offer diversification and support more efficient capital structures, but cross-defaults, covenant testing, tax, insolvency and regulatory issues make multi-asset and cross-border deals more challenging.
• Once assets are operational, refinancing into capital markets through term loans, private placements or securitizations can lower financing costs, though renewal risk, rating agency requirements and insolvency-remoteness concerns remain key hurdles.

Driven by rapid expansion from major AI developers and hyperscaler tenants (including Apple, Google and Microsoft), research anticipates 100 GW of new data center capacity coming online between 2026 and 2030¹. Meeting this increased demand will require USD3 trillion of investment, creating an unprecedented opportunity for market participants to deploy diversified liquidity structures across banks, public markets and private capital.

Financing digital infrastructure presents a unique challenge. It requires a financing model that accounts for hardware replacement cycles, renewable energy sourcing, sustainability compliance, revenue uncertainty, and significant capital outlay, all while maintaining predictable cash flows and returns. These complexities create hurdles for both developers and financiers in structuring suitable solutions.

Blending financing from acquisition to income generation

The foundation of any data center financing strategy begins with acquisition and construction debt. However, to achieve the most efficient structure over a data center project’s life, developers increasingly plan for how capital needs and risks will evolve, enabling a transition over time to lower-cost, more efficient debt.

Lenders have traditionally approached data center financing from a real estate perspective, underwriting credit based on the value of the land and physical property rather than operational performance (though tenor or pricing can be influenced by pre-letting arrangements). This type of financing is better suited to the acquisition stage or early development, where asset value is still secured by the underlying real estate.

As a data center project moves into a construction phase, developers look to raise non-recourse debt from banks and/or institutional investors. Here their financing strategy evolves into a blended real estate/infrastructure model, often combining a real estate loan with a development-phase or project financing transition.

Infrastructure financing is underwritten on the creditworthiness of tenants, quality, and risk profile of lease cash flows, the developer’s track record, and financeable construction arrangements with creditworthy counterparties. This blended approach mirrors the changing risk profile of the data center asset over time and consequently its bankability for banks and/or institutional investors.

The challenge is that the interests of construction, equipment and real estate lenders are distinct and frequently conflicting. Careful negotiation is needed to align competing positions—and broad due diligence is required to assess the bundle of rights needed to operate the asset and the viability of income-generating customer contracts. This results in a complex collateral and cash flow framework addressing lien priority, step-in and enforcement rights, pre- and post-completion waterfalls, and insolvency remoteness protections.

The classic “chicken and egg” problem illustrates this conflict: potential tenants require certainty that a site will deliver the required capacity (including grid connections, permits, and construction commencement) before committing to a contract.

Conversely, infrastructure lenders require signed customer contracts before committing capital, as these provide revenue certainty and reduce credit risk. This creates a capital gap that developers must bridge via equity or, increasingly, private credit and structured/securitization facilities in order to unlock large-scale infrastructure debt.

As competition for capital grows, lenders increasingly require a clear development narrative addressing asset diligence, tax and structuring issues within a complex and evolving regulatory environment. A lender’s desire to maintain supervision and step-in rights over material project documents must be balanced against the developer’s and hyperscaler’s need for operational flexibility. Developing an efficient funding structure is key to unlocking and sustaining growth at scale.

AI infrastructure financing

Until recently, most financings have focused on cloud data centers, the shell comprising land, building structure, and essential infrastructure such as cooling towers, grid connectivity, and security. The shell is then leased to cloud providers who install their own core components (servers, chips, and racks). This is the typical construct for hyperscale data centers.

In contrast, AI-focused data centers require the developer to own both shell and core and offer computing capacity (such as GPU-as-a-service, a model we explore in more detail here) directly to customers. The core assets significantly increase financing requirements, with their value almost four times higher than that of the shell.

This dynamic adds further complexity. Equipment financing introduces a different risk profile, with obsolescence being a primary concern; hardware typically has a maximum useful life of five years and will amortize fully or with a small tail value. The core assets may be separately financed, adding to discussions around security and enforcement rights and the mismatch between equipment lifecycle and long-dated customer contracts.

Single asset vs. multiple asset financing

After taking into account the issues outlined above, on a pure asset basis, the financing of a single data center is in principle relatively straightforward, revolving as it does around a single asset with a single set of cash flows. Financing a portfolio of assets introduces additional complexities in developing a suitable financing strategy. A portfolio financing offers risk diversification through multiple assets, staggered contract tenors, and lower renewal risk. However, terms may vary depending on portfolio strength and bankability. Financiers may require protections against defaulting assets, leading to challenging negotiations around cross-default provisions, isolation of defaulting assets for enforcement, and portfolio-level covenant testing.

For larger platforms, developers commonly adopt a bifurcated corporate structure separating development-phase assets from stable income-generating assets or use holdco-level financing structures that consolidate cash flows from multiple assets. Both approaches aim to optimize capital allocation, allowing operational flexibility while leveraging the stability of mature assets. Once a portfolio is sufficiently sizable, securitization, private placements, and other structured credit options become viable.

Regulatory fragmentation across Europe and the UK makes packaging cross-jurisdictional portfolios in a single transaction particularly challenging, given differing tax and insolvency regimes. Fortunately, given the relative size and value of data center assets, jurisdiction-by-jurisdiction financing remains viable. It remains to be seen whether master trust structures (which are commonly used in the more developed U.S. market to allow inclusion of additional assets in the collateral pool) could be adopted for cross-border European transactions.

Refinancing into the capital markets

Once a data center is operational and generating predictable lease revenue, the risk profile fundamentally changes. Construction risk is eliminated and initial construction debt is often replaced with permanent financing.

Takeout financing typically offers more favorable terms and lower borrowing costs, with the credit narrative looking through to the creditworthiness of hyperscaler tenants under long-term customer contracts. Options include broadly syndicated term loans, private placements, and, more recently, asset-backed securitizations in Europe and the UK.

There are, however, several hurdles to the development of long-term financing solutions, which remains a relatively nascent market.

Location is a key consideration. The European market is typically characterized by five primary hubs: Frankfurt, London, Amsterdam, Paris, and Dublin (FLAPD), due to their strategic locations and developed infrastructure.

These have been the jurisdictions from which data center securitizations have emerged in Europe, where asset scale and maturity have supported successful asset-backed security (ABS) execution. Secondary locations such as Berlin, Milan, Madrid, and Warsaw are beginning to emerge as local power connectivity and infrastructure improve, though it remains to be seen whether these markets will develop sufficiently to meet growing capital demands.

A further challenge is the mismatch between desired financing tenors and customer contract duration. Private placements often feature tenors of 20+ years, whereas hyperscale customer contracts are typically only ten years, introducing renewal risk and limiting the appeal of single-asset structures.

Larger portfolio financings offer risk diversification through staggered maturities. There is also a strong case for institutional investors accepting renewal risk where tenant creditworthiness is strong, extension options are embedded in contracts, and assets are located in favorable FLAPD markets.

Data center securitizations typically have an expected maturity of five years (that is, an anticipated repayment date (ARD)). There is growing debate about extending expected maturity to seven or ten years, with investors therefore being asked to accept renewal risk.

One further challenge with this structure is that failure to repay the transaction at expected maturity, while disincentivized with step-up coupons and accelerated amortization, means that hedging of the floating rate instruments preferred by the European investor base is difficult and expensive.

Another hurdle is the application of rating agency methodology, which requires the data center assets and related customer contracts to be owned by an insolvency-remote financing entity (or a subsidiary).

This is generally commercially undesirable, can involve adverse transfer tax consequences, and may be practically difficult due to restrictive national security regimes (such as the UK’s National Security and Investment Act.

Securitizations therefore need to be structured to ensure that financing and property-owning entities are, to the greatest extent possible, insolvency remote while remaining within the sponsor’s corporate group, resulting in less-efficient financing, complex tax structuring, careful negotiations on shared power and infrastructure contracts, and extensive discussions on enforcement rights.

Looking ahead

The data center financing market is rapidly maturing, presenting an unprecedented opportunity for banks, institutional investors, and private credit participants.

At the same time, the traditional lines between real estate and infrastructure loans, structured credit, and capital market solutions, are blurring. A coherent capital strategy therefore requires careful planning across these structures to align with distinct stages of a data center asset’s life cycle while optimizing the cost of capital in an increasingly competitive market.

Footnote

1. “2026 Global Data Center Outlook,” January 5, 2026, JLL.

The content of this article is intended to provide a general guide to the subject matter. Specialist advice should be sought about your specific circumstances.