Phil Wiltshire, Contracts Manager at Pensdown, explains why the success of data centre power infrastructure comes down to how well it is installed, tested and maintained on site.

Most people assume the digital world just works. A file uploads, a payment clears, a system responds, and it all feels instant. What they don’t see is the reality behind it – infrastructure that cannot fail, and built environments where there is no room for error. For me, the challenge is not just understanding how data centres are designed; it is delivering those designs on site. When installing complex electrical systems, often in live facilities, every connection, every sequence and every decision has to work first time.

The industry often talks about capacity, demand and growth, and rightly so. Global electricity consumption from data centres is estimated to have reached around 415 terawatt-hours in 2024, roughly 1.5% of global demand, according to the International Energy Agency, but those figures only tell part of the story. The real pressure is felt at project level, where electrical infrastructure must be installed, upgraded or expanded without compromising uptime, often within facilities that are already operational.

Designing power distribution that works in reality

On paper, power distribution in data centres is built around resilience. Dual utility feeds, segregated distribution paths and redundancy models such as N+1 or 2N are standard practice. In reality, implementing these systems on site introduces a different set of challenges.

Switchgear selection and layout, for example, is not just an engineering decision; it is a spatial and logistical one. Plant rooms are often constrained, particularly in retrofit or upgrade scenarios, and installing large switchgear assemblies while maintaining access, segregation and safety clearances requires careful coordination from the outset.

The same applies to cable routing. High-capacity power cables must be installed with consideration for thermal performance, electromagnetic interference and physical segregation. In live environments, routing becomes even more complex. Existing services, operational constraints and limited shutdown windows mean that installation is often phased, requiring temporary supplies and detailed sequencing to avoid disruption.

Uninterruptible power supplies are central to data centre resilience. They bridge the gap between grid failure and generator start-up, ensuring continuous power to critical loads. Industry data shows that UPS infrastructure represents a significant proportion of capital investment in data centre power systems, reflecting its importance in maintaining uptime. From a delivery perspective, however, integrating UPS systems is far from straightforward.

Battery systems, whether traditional VRLA or increasingly lithium-ion, require careful handling, installation and commissioning. Space, weight and ventilation all need to be considered, particularly in existing buildings. The interface between UPS systems and upstream switchgear must be precisely coordinated to ensure seamless transfer under fault conditions.

Testing is another critical phase. Load bank testing, system synchronisation and failover scenarios must be validated under controlled conditions. This requires detailed planning and close collaboration between contractors, manufacturers and commissioning teams. Any oversight at this stage can have serious implications once the facility is live.

Working in live environments

One of the biggest challenges in data centre delivery is working around live operations. Unlike new-build projects, where systems can be installed and tested before energisation, many data centre projects involve upgrades or expansions within operational facilities. This changes the risk profile entirely.

Shutdowns are tightly controlled and often limited to very short windows. In some cases, work must be carried out while systems remain energised, requiring strict safety procedures and experienced personnel. Coordination becomes critical, not just between trades, but with the client’s operations team.

Every activity must be planned in detail. Method statements, risk assessments and sequencing are not simply paperwork; they are essential tools for ensuring that work can be carried out safely without impacting service.

This is because electrical infrastructure in data centres does not operate in isolation. It is closely linked to mechanical systems, particularly cooling. As computing density increases, so does the demand on cooling systems. According to the International Energy Agency, cooling can account for a significant proportion of total data centre energy use, often approaching 30 to 40% in some facilities. This creates a direct relationship between electrical and mechanical design.

From a delivery perspective, this means close coordination is essential. Power supplies to cooling equipment must be aligned with redundancy strategies, ensuring that failure in one system does not compromise the other. Control systems, monitoring and integration all need to be considered.

In practice, this requires constant communication between electrical and mechanical teams, particularly during installation and commissioning. Misalignment at this stage can lead to inefficiencies or, in the worst cases, operational risk.

Commissioning is where design intent meets operational reality. In data centres, this process is rigorous. Systems must be tested under a range of conditions, including simulated failures, to ensure they perform as expected. This includes verifying redundancy, load sharing and transfer sequences.

Documentation plays a critical role. Detailed records of installation, testing and system configuration are essential not only for compliance, but also for ongoing operation and maintenance. Facilities teams rely on accurate information to manage and troubleshoot systems over time.

From experience, the quality of commissioning and documentation often determines the long-term success of a project. It is not enough for systems to work on day one; they must be understood and maintainable for years to come.

Building for growth without disruption

Growth is a defining feature of modern data centres. Facilities are designed to expand as demand grows, often in phases. This has significant implications for electrical design and delivery. Infrastructure must be installed in a way that allows future capacity to be added without major disruption. This includes selecting switchgear and UPS systems that can be expanded, as well as designing distribution networks that can accommodate increased load.

In practice, this means thinking beyond the immediate project. Cable routes, containment systems and plant layouts must all consider future requirements. Failure to do so can lead to costly and disruptive retrofits later.

Equally important, there is a tendency to view data centre infrastructure through the lens of technology and innovation. While those elements are important, the success of any project ultimately comes down to delivery.

From power distribution and UPS integration to cable routing and commissioning, the challenges are practical, detailed and often constrained by real-world conditions. Working in live environments, coordinating with other systems and ensuring compliance with stringent standards all add layers of complexity.

In my experience, resilience is not just designed; it is built, tested and proven on site. It comes from understanding how systems interact, how they are installed and how they will be maintained over time.

Data centres may be invisible to most people, but the infrastructure behind them is anything but simple. As demand continues to grow, driven by cloud services, AI and digital transformation, the pressure on electrical systems will only increase. Meeting that demand requires more than capacity; it requires infrastructure that can be delivered reliably, safely and with future growth in mind.

Because, in the end, resilience is not defined by design alone. It is defined by how well that design is delivered.