Globally, project finance structures are increasingly being used to bring data centre projects to market. Large-scale developments in Europe, Asia-Pacific and the Americas have demonstrated that the asset class can support non-recourse or limited-recourse financing where the right fundamentals are in place. The question to consider, then, is whether data centres are project-financeable in the way power plants and toll roads are. The answer is yes – but only if the structure honestly confronts the following three factors that influence the ways in which data centres differ from classic infrastructure.

1. The offtake problem: Hyperscale vs co-location

Project finance is, at its core, lending against contracted cash flows housed in a special purpose vehicle, with limited or no recourse to sponsors. In a power deal, the power purchase agreement (PPA) is the credit. In a data centre, the equivalent is the customer contract – and not all customer contracts are created equal.

A hyperscale pre-let or anchor tenancy with a global cloud provider can look remarkably like a PPA: long tenor, creditworthy counterparty or take-or-pay-style capacity commitments. Lenders can and do underwrite against these, and direct agreements with anchor tenants (including step-in rights on default) are becoming a standard part of the security package.

Retail co-location is a different animal. Contracts typically run three to five years against debt tenors of 10 to 15 years. That mismatch pushes deals toward hybrid structures: a project finance core sized against contracted anchor revenue, with an uncontracted “merchant” layer financed through equity, mezzanine or holdco debt. The multi-tranche structures we are now seeing in East African deals reflect exactly this layered risk allocation.

2. The power problem: Every African data centre is also an energy project

In mature markets, grid reliability is assumed. In most African markets, it cannot be. Facilities routinely embed captive generation, solar and battery storage – meaning the financing must underwrite what is effectively a small independent power producer inside the project perimeter. That brings a second regulatory workstream (in Kenya, Energy and Petroleum Regulatory Authority licensing for captive power and any wheeling arrangements), a second engineering, procurement and construction (EPC) and operations and maintenance (O&M) contract structure, and a second set of technology and fuel-supply risks.

The upside is that renewable self-generation is increasingly a bankability advantage rather than a burden. It hedges grid risk, reduces long-run operational expenditure and opens the door to green loan and sustainability-linked pricing – a meaningful lever when development finance institutions (DFIs), which have deployed an estimated USD 1,5–2 billion into African data centres since 2016, sit at the heart of most syndicates.

3. The currency problem: Dollar debt, shilling revenue

Data centre capital expenditure (CAPEX) is overwhelmingly dollar denominated. Revenue is often a blend of hyperscale contracts priced in USD, domestic enterprise and government co-location in local currency. Lenders will interrogate the currency composition of the contracted revenue stack, and sponsors should expect requirements around offshore collection accounts, hedging programmes, or natural hedges through USD-linked pricing. Where government or parastatal offtake features, the analysis extends to convertibility and transferability risk – familiar territory for anyone who has closed an African PPA.

What the security package looks like

The emerging African data centre financing borrows heavily from the power finance toolkit: share pledges over the special purpose vehicle, all-asset debentures, assignment of material contracts (customer agreements, EPC, O&M, power supply), direct agreements with anchor tenants and key contractors, and insurance assignments. Two data-centre-specific additions matter. First, lenders increasingly want step-in mechanics that preserve operational continuity – a data centre in default is worthless to a lender if customers migrate workloads out during enforcement. Second, land and title diligence carries elevated weight, because the asset is immovable, CAPEX-intensive and often built on long leaseholds. Where DFIs and commercial lenders participate together, intercreditor arrangements and subordination mechanics will also feature prominently in the documentation.

The regulatory overlay is now a demand driver

Over 40 African countries have enacted data protection legislation – including South Africa’s Protection of Personal Information Act 4 of 2013, Kenya’s Data Protection Act, and Nigeria’s Nigeria Data Protection Regulation – and data localisation requirements are converting legal compliance into commercial demand for in-country capacity. Governments are treating data centres as critical national infrastructure. For investors, this cuts both ways: sovereignty rules underpin the demand case, but they also mean licensing, change-of-control approvals, and in some markets local ownership considerations must be mapped early – ideally before the term sheet, not after.

The bottom line

Data centres sit somewhere between real estate, infrastructure and technology – and the financing structures reflect that hybridity. The deals being done in Nairobi, Lagos and Johannesburg today are neither pure corporate loans nor textbook project financings. They are layered structures: contracted-revenue project debt at the core, flexible capital around it, DFI participation for tenor and political risk comfort, and a security package built for operational continuity.

For sponsors, the practical takeaway is to structure for bankability from day one: anchor tenancy strategy, power solution, land tenure, and licensing are financing questions, not just operational ones. For lenders, the opportunity is significant – but the underwriting discipline of power finance travels better here than the assumptions of European real estate lending.