The choice of how an organisation accesses data center capacity — colocation, build-to-suit, or a dedicated sovereign platform — has significant implications for capital efficiency, control, and risk that go well beyond a simple lease-versus-own decision. As AI infrastructure demand grows more sophisticated, so has the range of structures available to satisfy it.
Colocation Remains the Most Capital-Efficient Entry Point
Colocation — leasing space, power, and cooling within a shared, multi-tenant facility — remains the most capital-efficient way for organisations to access data center capacity without taking on development risk directly. This structure works well for workloads that do not require highly bespoke power density, security, or governance arrangements, and it allows occupiers to scale capacity up or down with greater flexibility than owning dedicated infrastructure typically allows.
Its limitations become more apparent for organisations with extreme density requirements, stringent sovereignty or security needs, or AI infrastructure ambitions large enough that dedicated, purpose-built capacity becomes more cost-effective than continually leasing premium colocation space at scale.
Build-to-Suit Bridges the Gap
- Build-to-suit arrangements allow an occupier to specify a facility's design — power density, cooling technology, security architecture — without taking on full development and ownership risk themselves
- This structure typically involves a developer or specialist operator constructing and sometimes continuing to own the facility, with the occupier committed to a long-term lease that underwrites the development financing
- Build-to-suit is particularly well suited to large AI infrastructure tenants with specific, well-defined technical requirements that standard colocation products cannot efficiently satisfy
Sovereign Platforms Represent the Most Bespoke End of the Spectrum
For governments and organisations with stringent sovereignty, security, or strategic control requirements, fully sovereign infrastructure platforms — owned, governed, and often operated entirely within structures designed to satisfy specific jurisdictional and control requirements — represent the most bespoke and capital-intensive option, but also the only structure capable of satisfying the most demanding sovereignty requirements discussed elsewhere in our analysis of European strategic autonomy and data residency.
The right ownership structure is not the one that maximises control or minimises cost in isolation — it is the one that matches an organisation's actual risk tolerance, technical requirements and capital strategy.
Hybrid Strategies Are Increasingly Common
Many sophisticated organisations do not commit exclusively to a single structure across their entire infrastructure footprint. A common pattern involves using colocation for more standard, lower-density workloads while pursuing build-to-suit or sovereign platform structures specifically for the highest-density AI training capacity or the most sovereignty-sensitive workloads — allowing capital and control to be allocated where they add the most value, rather than applying a single structural template uniformly across a diverse infrastructure portfolio.
Choosing the Right Structure for Your Strategy
DATAPERT advises clients on evaluating these structures against their specific technical, governance, and capital objectives as part of our investment advisory services. Explore our broader data center development capabilities or start a project to discuss the right infrastructure ownership structure for your organisation.
