As organisations across Europe accelerate AI adoption, Keiran O’Gorman, Technical Lead, Major Projects, Eland Cables, says the demands placed on data centre infrastructure are forcing CIOs and operators to rethink long-held assumptions around power, cooling, resilience and scalability.
Across Europe, CIOs are shaping ambitious AI roadmaps designed to unlock productivity, new customer experiences and enhanced digital services. Yet beneath the optimism lies an important inflection point, one that gives organisations a chance to rethink the infrastructure supporting their AI ambitions. Building and network designs agreed only a few years ago can feel increasingly stretched by the electrical, thermal and operational demands associated with modern AI workloads.
Encouragingly though, the data centre industry now has more insight and real-world performance data than ever before. Organisations that recognise this early are already positioning themselves to build stronger, more adaptable platforms for AI.
Working across hyperscale and enterprise projects gives me visibility of the recurring challenges that often emerge long before equipment is installed. These are not dramatic or unusual failures, but patterns that tend to surface during design reviews, procurement cycles, cable specification discussions and late-stage commissioning; moments where long-established assumptions are tested against evolving requirements.
How AI is reshaping long-held assumptions
For more than a decade, data centre design evolved in broadly predictable ways. Power densities rose slowly, cooling technologies matured and engineers could rely on established models that translated well across multiple European regions. AI has rewritten those expectations.
Linear, stable power draws have given way to highly variable loads. Cooling systems built for conventional server behaviour now contend with concentrated, rapidly fluctuating heat. Even the physical space required for cables, containment and switchgear is narrowing as operators squeeze more capability into buildings never intended for this level of density.
These conditions differ from site to site, but each presents an opportunity to design smarter and build more resilient infrastructure. In southern Europe and the deserts of the US, operators wrestle with extreme heat and high UV exposure, forcing choices around cables, materials and fire classifications that would be irrelevant in cooler climates. In the Nordics, the issue flips. Bitter winter temperatures and persistent moisture create a separate set of constraints.
Across Europe, regulations such as CPR classifications and building standards are harmonised in theory, yet interpreted through different national lenses in practice.
Rather than being barriers, these regional differences offer valuable design insight. The more organisations learn from these conditions, the more they can strengthen their infrastructure elsewhere. What once looked like fragmentation is increasingly a source of design maturity.
Capacity now has a different meaning
Many organisations still view capacity through a narrow lens: floor space, available power and cooling headroom. AI is widening that definition. Today, land availability, planning processes, grid access and skilled labour all influence whether an AI project progresses smoothly.
Across Europe, national grids are under growing pressure. Some regions have paused new data centre developments until networks can reliably absorb additional demand. As a result, planning authorities now expect clarity around power sourcing, renewable integration and long-term grid impact; expectations that are reshaping design decisions earlier in the cycle.
Labour dynamics add another dimension. Large builds require thousands of skilled tradespeople onsite at peak, often in locations where local workforces cannot meet demand. Contractors rotate teams across borders and the logistics of doing so now factor into programme certainty just as much as equipment lead times.
Yet this intense, transient labour requirement stands in stark contrast to the comparatively low operational workforce needed once the build is complete.
When viewed together, these pressures reshape what capacity really means. Capacity extends beyond racks and megawatts. It encompasses grid infrastructure, regional workforce availability and the readiness of local authorities to support expansion. Once understood, these factors become powerful tools for anticipating and mitigating risk.
Adapting designs as AI requirements evolve
One of the clearest signs of AI’s impact is the increase in design revisions during construction. Operators who committed to a particular cooling or electrical strategy early in the programme often revisit those decisions as hardware requirements evolve or as tenants introduce new power demands.
These revisions can be disruptive, but they also demonstrate how quickly organisations can adapt when required. Systems are being upgraded ahead of first deployment and in many cases this proactive approach is improving long-term flexibility. What looks like rework is, in reality, the industry learning in real time.
For CIOs, this trend is less a threat and more a signal that AI strategies must be anchored in infrastructure reality. When expectations outpace what a facility can deliver, projects can encounter avoidable delays and additional costs.
Recognising these limits early allows organisations to plan upgrades on their own terms, rather than reacting under pressure.
Rethinking AI strategy through an infrastructure lens
Enterprises do not need to scale back their AI ambition; they simply need to ground it in the realities of the environments where these workloads operate. The reward is a deeper, more accurate understanding of what AI requires and the chance to build platforms that remain robust as technology evolves.
Well-grounded AI strategies begin with a realistic appraisal of the site conditions, regulatory frameworks, material behaviours and grid dependencies that shape each location. They acknowledge that a facility in Frankfurt does not operate exactly like one in Madrid, Helsinki or Dublin, even when built to similar specifications.
They recognise that planning cycles and construction cycles do not move at the pace of digital adoption, meaning flexibility must be designed in from the start.
Early collaboration with partners who understand these pressures is essential. Cable and power distribution design is a strategic lever, not a technical afterthought, and it directly determines power density, cooling strategy and long-term adaptability. When addressed early, these fundamentals create room for the facility and the AI strategy to grow.
A more grounded path forward
AI will continue to influence every aspect of digital operations and data centres will remain the critical infrastructure enabling that change. But ambition alone cannot deliver the scale organisations are aiming for.
Success will come from those who understand that AI readiness is built not only on compute and algorithms, but on power networks, planning frameworks, skilled labour and the materials and components that form the backbone of reliable operation.
Infrastructure moves at the pace of real-world conditions, but when those conditions are understood early, organisations can shape solutions rather than react to them. These forces may move slowly, but they offer clarity and predictability — the foundations of well-planned AI transformation.
The future of AI will be built on extraordinary technical breakthroughs, but it will rely just as much on thoughtful design, early planning and the expertise that keeps infrastructure aligned with ambition. For organisations willing to meet this moment with foresight and the right partnerships, the path ahead is not just navigable, it is full of opportunity.
