Planning considerations for data centre development

Revenue streams vary across the different types of data centres. For example:

• Clocation data centres generate direct, external revenue by leasing space, power and connectivity to multiple customers. Income is typically contracted and recurring, priced per kW, rack or other variables. This creates an annuity-style cash flow across several tenants (diversified cash flow stream).
• Hyperscale data centres, by contrast, rarely lease to third parties; instead, they build infrastructure to support their own cloud platforms (e.g. AWS, Google). Revenue is generated upstream through cloud services (AI workloads) rather than facility leasing. 

At the design phase, emissions are addressed by:

• Embedding sustainability into core engineering decisions, including site selection in regions with low-carbon grids or access to renewable energy, supported by long-term power purchase agreements (PPAs).
• Prioritising energy-efficient layouts and infrastructure, targeting low Power Usage Effectiveness (PUE) through optimised electrical systems and adoption of increasingly advanced cooling architectures such as liquid or direct-to-chip cooling
• Focusing on life-cycle assessment (LCA) to minimise embodied carbon through material selection and construction methods, i.e., modular
• Considering whole of life of design principles

During operations, the focus shifts to:

• Monitoring and tracking energy/water consumption and heat loads and decarbonising power supply
• Heat-reuse strategies to further reduce energy demand and waste
• Developing management and maintenance systems to extend the life of the hardware and building fabric

Power remains the dominant constraint. However, permitting and water are emerging as equally material risks, particularly environmental approvals.

Most clients still view inputs from separate streams (utilities, planners, regulators), but the market is shifting toward a more integrated “power + permission + sustainability” approach at the earliest stages of site selection, as fragmentation increasingly leads to failed or delayed projects.

However, the fourth risk is social licence. Coordinated and focused community action can stop a project, regardless of the other three constraints. This is not an emerging risk; it is live and real.

• Hyperscale: Power-led and strategic. Site selection is driven by securing large-scale, long-term power, land and ability to expand capacity 100MW+ multi-phase investment.

• Sub-hyperscale (<20MW): Demand-led and flexible. Site selection balances connectivity, speed to market and existing infrastructure, with quicker, more iterative feasibility.

Bottom line: hyperscale is about locking in future capacity with high certainty, while sub-hyperscale is about delivering smaller, faster and more adaptable solutions.

Tenants typically issue LOIs very early, once the initial feasibility (especially power and permitting viability) is understood, but well before full design, approvals or construction.

This early engagement is directly responsible for increasing the speed to market of a development. As developers lock in tenants based on development programs, very few developments are speculative.

Yes, this is already happening, particularly in Europe, and is gaining real traction, but it’s typically done at a district scale rather than direct building co location.

The issue in many jurisdictions, however, is not the tech but rather the myriad of regulation and different agencies required to provide approval that is restricting adoption.

It depends. The construction phase of a data centre creates significant employment and delivers real, economic benefits to local businesses. However, once operational, data centres are relatively low employment assets, so “jobs” alone is often not a compelling long term argument for approval.

Increasingly, authorities are looking beyond employment to a broader value proposition, such as infrastructure investment, energy integration (e.g., heat reuse), local economic uplift and sustainability outcomes.

Employment helps, but on its own it is unlikely to secure approval. Projects need to demonstrate a wider, enduring community benefit.

Data centres are increasingly being treated as core infrastructure inputs into energy, water and urban system planning, rather than standalone developments. This means aligning site selection and delivery with grid/water planning and regional industrial strategies from the outset.

In Finland, hyperscale data centres are embedded into district energy planning, where waste heat is captured and fed into municipal heating networks, supplying a large share of residential heating demand. This is coordinated with utilities and local authorities as part of a system-level approach to both power and heat infrastructure rather than a standalone facility decision.

If we take this as two questions. Firstly, data centre development. We know the constraints in relation to water and energy, which relate directly to the need for conditioned space to manage heat loads. Research into how this heat load can be reduced, both economically and environmentally, with a corresponding reduction in water and power usage is a clear area where collaboration between academia and industry would be highly beneficial.

In relation to AI, the challenge is different. The scale of compute, data and investment required limits academia’s ability to contribute without the support of industry. The gap is therefore not capability, but access to compute, operational data and resources.

Academia can help solve the system-level challenges, particularly in energy, cooling and optimisation, but it needs to be more tightly integrated with industry to do so at meaningful scale.

At a practical level, innovation is what allows the industry to keep growing and, in some cases, be viable. For example, cooling alone can account for a significant share of energy use, while advances like liquid and direct-to-chip cooling materially reduce both power and water consumption. 

Research underpins sustainability and social licence. Tools like lifecycle analysis are now being used to design data centres that reduce carbon, energy and water use upfront, rather than retrospectively. 

Without ongoing research and innovation, the sector would hit hard limits on power, cooling and community acceptance.

Yes, but it is possibly time to rethink data centres as integrated pieces of infrastructure.

What this means in practice is that we treat data centres as part of a coordinated system solution, co-located with renewable generation, integrated with storage, designed to reuse heat and aligned with long-term grid and water planning. In that model, they can help enable investment in new infrastructure, particularly renewable energy.

So yes, we can keep up, but only if we shift from a project-by-project mindset to a system-planned, infrastructure-led approach.

The primary constraints are access to scalable, reliable power, water and fibre connectivity. These are the non negotiables that ultimately determine whether a region is viable.

Beyond that, availability of suitable land is critical, particularly sites that are free from natural risks, such as seismic activity, and those that can support long-term expansion.

Equally important is the local community perspective. Social licence is now a defining factor. Regions that can demonstrate clear community alignment, transparent planning and tangible local benefits are far more likely to attract and sustain data centre development.

At the front end, there is a much stronger use of integrated planning and modelling tools that allow developers, utilities and governments to test scenarios around power, water and load growth before projects are committed.

From a market perspective, energy platforms and PPA structuring tools help connect data centres with renewable developers, essentially aligning demand and supply in a more coordinated way.

And importantly, we’re starting to see shared frameworks and standards emerge, whether that’s energy performance frameworks, ESG reporting or planning guidelines, which give investors and regulators a common language to work from.

Communities should be shown a clear, transparent picture of impact in real, practical terms. This includes visuals describing what a facility will look like and how it will operate day to day (noise, traffic, scale), alongside a simple, evidence-based explanation of power and water use and how those impacts are being managed. 

Equally important is explanation of why the site was chosen, how it fits into the broader energy and digital system and what commitments are in place to monitor and manage impacts over time. In essence, moving beyond technical detail to tangible non-technical details.

Supply chain is becoming a core part of integrated planning. With long lead times on critical equipment — power systems, generators, cooling infrastructure and even grid components — the ability to deliver a data centre is now directly tied to how early and effectively the supply chain is engaged. 

In practice, that means aligning site selection, power strategy and procurement upfront, often securing key equipment and capacity at the same time as land or power agreements. This in turn changes the deliverable mindset during the design phase and drives greater standardisation / modular design, allowing components to be deployed faster and at scale.

In simple terms, yes, but there are a couple of practical challenges to resolve. 

The first is the physical connection, i.e., getting the piping and infrastructure in place to transfer heat between uses in a reliable and cost-effective way, and what agencies are involved in providing approvals and access if public space is to be traversed. 

The second is mismatch of demand. A data centre operates 24/7, producing a constant heat load, whereas the end user (e.g., residential or commercial heating) is often variable. 

The question then becomes how to manage periods where demand is low, typically requiring storage, buffering or integration with a broader energy system to make it viable.

The first is modular design and construction. Moving toward highly standardised, prefabricated and scalable solutions, i.e., repeatable infrastructure rather than bespoke builds.

Secondly, technologies that directly reduce heat loads, which correspondingly reduce both cooling energy and water consumption.