Recently, Eddie Beer, Director at IN2 Engineering, had the pleasure of joining an expert panel discussion focusing on the risks, opportunities and practical delivery challenges facing the data centre sector.
Joining him on the day were experts in various fields involved in the management and delivery of data centres, including Paul Gibbons, CEO at Decipher; Mark Rimmer, Director at Decipher; Richard O’Brien, CEO at Merlin; Syed Muzaffar, Director at Decipher; and David Banks, Partner at Hill Dickinson.
One of the key discussion topics was what makes data centre projects so uniquely complex compared with other building types.
Drawing on the panel discussion, Eddie outlines below five key factors that make data centres uniquely complex: from the dominance of MEP and technology infrastructure, to the pace of digital change, the consequence of failure, commissioning intensity and the growing challenge of power availability.
They are MEP and technology-led buildings
In most buildings, the MEP and IT infrastructure serve the building. In a data centre, the building serves the MEP and IT infrastructure. These projects are technology-led developments.
Power, cooling, resilience, controls, security and connectivity are not secondary considerations; they are the core of the project. The architecture, structure, spatial planning and programme all need to respond to the technical requirements, rather than the other way around.
Data centres also contain a huge volume of plant and equipment, much of it duplicated or arranged in resilient topologies to minimise single points of failure.
With this extent of technology, comes complexity: more systems, more redundancy, more interfaces, more controls, more testing and more things that need to work together.
Technology moves faster than construction
The pace of technological change is another major challenge. A project may take years to plan, design, procure, build and commission, but the underlying IT and MEP technology can evolve several times within that same period.
The shift from CPU-led compute to GPU-heavy AI workloads is a clear example. Higher rack densities drive more power demand, more heat, different cooling requirements and, in many cases, a move from traditional air cooling toward hybrid or liquid cooling solutions.
This means data centre design needs flexibility, but that flexibility must be deliberate, scoped and managed. Vague “future-proofing” can quickly become uncontrolled change.
The consequence of failure is high
In a typical commercial building, poor MEP performance may cause discomfort, inefficiency or tenant complaints. Clearly none of these are ideal, but in a data centre, loss of power or cooling can mean service interruption, equipment risk, reputational damage and of course significant financial impact. With some hyperscale data centres generating revenue in the tens of millions per day, downtime risk is something that must be mitigated at all costs.
This is why resilience is central to data centre design.
The question is not simply whether there is enough plant, but whether the topology can survive maintenance, faults and abnormal operating conditions. Even a relatively small component failure can cascade if the system architecture, controls or commissioning are weak.
Commissioning is a project in itself
Commissioning in data centres is significantly more intensive than in many other sectors. It is the point where the design proves whether it can keep its promises.
The process typically moves from factory acceptance testing and site acceptance testing through to functional testing and final integrated systems testing. At the final stage, systems are tested under realistic load, transfer and failure scenarios.
This process takes time and must be properly allowed for in the programme. Compressing commissioning to recover earlier delay is rarely the solution.
Power: the viability test for data centres
No data centre discussion today is complete without mentioning power. Large schemes may require tens or even hundreds of megawatts, at a time when the national grid is also being asked to support electrification of heat, transport and industry.
In most sectors, site viability is judged by familiar factors: location, planning, land value, access, market demand, and construction cost. In data centres, those factors still matter, but they are no longer enough.
The critical question is whether the site can secure sufficient power, at the right voltage, within a feasible timeframe. Without that, the site is not truly viable.
Capacity, voltage, connection date, reinforcement, phasing and resilience all affect site selection, planning, cost and programme. In constrained locations, connection timelines can become commercially decisive.
Reform is underway in the UK to move away from a simple “first come, first served” queue and toward a model that prioritises projects which are ready, viable and strategically needed. That should help serious projects progress, but it will not remove the fundamental challenge: the physical grid still needs capacity, substations, transmission, distribution, transformers and time.
The direction of reform is encouraging, but the industry will ultimately judge it by one measure: whether viable projects actually get energised within feasible timeframes.
The opportunity for UK and European data centre development is significant, but ultimately if the sector is to move forward at the pace required, grid infrastructure must keep up with industry ambition.