Power Before Everything: What the Data Centre Boom Demands from Electrical Engineering

‍

by Graham Robson | Head of Electrical Engineering

‍

Datacentres have become one of the defining infrastructure stories of our generation. If you work anywhere near power systems design, grid connections,or electrical engineering, you will have noticed it. The demand isn’t just growing, it’s accelerating in ways that are fundamentally reshaping what engineers need to know, and what clients need to ask for.

I’ve spent my career working in some of the most demanding electrical environmentsin the world - offshore oil and gas, where the consequence of getting it wrong isn’t a service interruption but something far worse. The discipline that environment instils has a direct and, I would argue, underappreciated relevance to what the data centre sector needs right now. In this piece, I want to share my perspective on why the electrical engineering challenge at the heart of these projects is bigger than most people appreciate and why getting it rightat the front end is everything.

‍

The Market: A Grid That Was Never Built for This

The single biggest pressure I see in the data centre sector right now is the sheer scale of power demand - and the grid's struggle to support it.

We’re not talking about incremental increases. We’re talking about facilities that require enormous, sustained power connections and a distribution network that wasn’t designed with this in mind. Distribution Network Operators (DNOs) are increasingly struggling to offer the connection sizes that data centre projects require.

When you apply for a DNO licence, you submit a number. That number is your ceiling. Get that number wrong - underestimate it or fail to plan for growth - and you face a fundamental constraint that is extremely difficult and expensive to resolve mid-project.

This is already driving decisions about where data centres get built. There’s a reason we’re seeing serious interest in Scotland and the north of England as locations. Scottish Enterprise, for example, has been actively exploring how data centres can be sited closer to where renewable energy is actually generated - wind power in particular. Right now, the UK Government is spending significant sums turning turbines off because demand isn’t in the right places.The idea of co-locating data centres with renewable generation sites is not a fringe concept – it’s a logical response to a genuine grid constraint problem.

And then there’s the cooling question. The further north you go, the more ambient conditions work in your favour for managing the heat load these facilities generate. These things aren’t coincidences but the engineering logic of the market playing out in real time.

What’s driving all of this demand? It’s easy to point to AI chatbots and search, and they are part of it. But the picture is much broader. The explosion in high-resolution computational rendering, the data storage requirements of modern military operations (every drone flying over a conflict zone is capturing high-definition footage that needs to be transmitted and stored somewhere), the growth of cloud-based enterprise computing, autonomous systems,simulation… all of it feeds the same beast. The data centre boom isn’t a technology trend… it’s more of a permanent feature of modern infrastructure.

Where Projects Go Wrong: Front-End Failure Is Almost Always the Culprit

In my view - and this is consistent with what I’ve seen across complex electrical projects in general - the most expensive mistakes are almost never made during construction. They’re made, or rather set in motion, much earlier: in the planning and design definition phase.

Ill preparation is the root cause. Not doing the due diligence upfront. Entering detailed design - or worse, procurement and construction - without having properly modelled and proven your power system architecture creates a cascade of downstream problems that get progressively more expensive to fix.

The DNO connection issue I mentioned earlier is a perfect example of this. If your original power budget was wrong, or if the load growth inherent in a facility of this scale was not properly accounted for from the outset, then the changes forced on you once the project is underway can be enormously disruptive. You cannot simply renegotiate a grid connection in the same way you might value-engineer a structural element. The lead times and regulatory processes involved mean that getting this wrong early costs you time, money, and potentially the viability of the project at it’s intended scale.

The lesson is straightforward: the electrical design process for a data centre must begin with a rigorous, modelled power system study, not with a layout drawing and a guess at demand.

‍

What Good Looks Like: Model It First, Then Build It

If I were advising an EPC procurement lead on what to ask for from an electrical engineering subcontractor on a data centre programme, this is what I would tell them:

Demand proper power system modelling from day one.

Before any detailed design work begins, the distribution network should be built in a power system simulation tool, like ETAP, and fully proven. That model should demonstrate that the DNO connection size is appropriate for the facility's actual requirement, not a best estimate. It should run load flow simulations for all operating scenarios, and it should size all major equipment -transformers, switchgear, protection systems - based on those proven results rather than rules of thumb.

What that modelling gives you is something invaluable which is a complete,simulation-proven picture of major equipment sizing, protection coordination, and safety systems all the way through the infrastructure of the facility. It’s the due diligence step that validates every major decision downstream.

Once that work has been done properly, something interesting happens: the rest of the electrical design becomes highly repeatable. Data centres, for all their complexity, have a level of architectural consistency that means once you’ve proven the design concept, the detailed design that follows is disciplined, systematic work. You’re not solving new problems on every drawing because you’re applying a proven solution at scale. That is efficient, cost-effective, and reduces the risk of error in detailed design.

What this tells EPC procurement leads is simple: front-load the engineering investment. The value delivered by rigorous power system design at concept and FEED stage far outweighs the cost - and protects the project against the much larger costs that come from getting it wrong.

Why Our Background Matters: Uptime Is Uptime

People may ask why an engineering firm with deep roots in offshore oil and gas is wellplaced to talk about data centres. I think the answer is obvious once you understand what both environments actually demand from electrical engineering.

Offshore,the consequence of electrical failure isn’t merely an inconvenience. It can be a safety emergency. Every electrical system we design is engineered for uptime,for resilience, and for the ability to manage failure scenarios without catastrophic outcome. You learn to treat power continuity as a non-negotiable engineering requirement, not an aspiration.

Datacentres require the same mindset. The stakes are different in that it’s computing uptime rather than personnel safety, but the engineering discipline required is directly comparable. A data centre's grid connection is its lifeline. If it fails, the facility fails. And a facility operating at this scale, serving AI workloads, enterprise systems, or real-time data processing, simply cannot afford that.

The answer to that vulnerability is a capable microgrid. For a facility of meaningful scale, that means a combination of renewables - wind and solar are the obvious starting points - backed by serious battery energy storage capacity. And when I say serious, I mean it: we are talking about battery storage at a scale that is genuinely hard to visualise until you are standing in front of it. Multiple fields of storage infrastructure, potentially with wind turbines and solar arrays integrated around it. That is not a supplementary system - it is a core piece of the facility's power architecture.

To illustrate the scale of what is being built: I am aware of a Microsoft facility that procured from a portable generation manufacturer circa 18 off 1500kVA generator sets, had them set up for a significant run time test in the company car park, such is the scale of project and level of investment required for the uptime assurance. That is not an edge case. That is the standard of resilience validation that hyperscale operator's demand. The engineering required to design, specify, and integrate power systems at that scale is not entry-level work.

Offshore engineering gives you a working knowledge of exactly these kinds of systems - portable and standby generation, uninterruptible power supply design, battery energy storage integration, and the protection coordination that holds it all together under fault conditions. That experience is directly transferable. And combined with the power system modelling capability we have built, it puts us in a position to engage meaningfully with the hardest parts of data centre electrical design.

‍

A Final Thought

The data centre sector is moving fast, and the companies that succeed in it will be the ones who treat electrical engineering as a first-order discipline rather than a downstream deliverable.

We know what is required. We understand power systems design, grid constraints, resilience engineering, and the uptime demands of mission-critical infrastructure. And we are actively looking to work with developers and EPCs who want an engineering partner that brings that depth to the front end of their data centre projects - where it makes the most difference.

Graham Robson is Head of Electrical Engineering. He has extensive experience in power system design for offshore and industrial environments and is leading the company's data centre electrical engineering capability development.

‍