AI infrastructure at a crossroads: why holistic data center design can’t wait

As artificial intelligence reshapes global industries, data center infrastructure has become the frontline of digital transformation.

Yet as operators rush to meet rising AI demands, they’re also facing growing geopolitical and economic headwinds. Global trade restrictions, skills shortages and tightening regulations are all adding complexity and cost to an already stretched ecosystem.

According to recent research, 69% of data center operators cite US tariffs and geopolitical instability as key drivers of delays and rising build costs, while 40% warn they are running out of viable sites to build AI-ready facilities.

In this high-stakes environment, having the right AI infrastructure has become a strategic imperative.

The scale and speed of AI growth require more than incremental change. Operators must embrace a holistic approach to design that ensures infrastructure is flexible, scalable and future ready.

Without it, they risk building sites that degrade quickly, require costly retrofits or fail to meet customer expectations.

AI surge, infrastructure strain

The arrival of large-scale AI workloads has upended traditional infrastructure planning. In the past year alone, 92% of data center operators have seen an increase in demand for AI capacity, with an average rise of 42%.

These workloads require far more power, cooling and network throughput than traditional applications, placing extreme stress on legacy designs.

Even more concerning, 64% of operators say demand has exceeded expectations, and 58% believe AI will shorten the lifespan of their existing facilities. In response, 74% are now rethinking their infrastructure strategies, adjusting plans for power, cooling, site selection and architectural flexibility.

Some operators are investing in greenfield developments purpose-built for AI. Others are retrofitting existing sites. Many are doing both.

However, progress remains uneven. In Ireland, for example, 12% of operators have paused AI infrastructure projects due to national restrictions on new developments driven by energy consumption concerns. In comparison, only 5% have done so in the UK, and just 2% in the Nordics.

Future success won’t come from raw capacity alone. It will depend on adaptability; data centers that can evolve alongside shifting workloads without compromising on energy efficiency or resilience, or sustainability.

The human factor

While AI’s technical demands are considerable, the true limiting factor may be people. Nearly 80% of operators say that skills shortages will delay their data center projects, and 45% report difficulty attracting talent with AI-specific expertise.

This challenge extends beyond internal teams. 28% of operators are struggling to find qualified partners for complex AI deployments. A further 18% cite a lack of AI-aligned design expertise. And as projects accelerate, 16% are seeing growing health and safety risks on site.

As infrastructure complexity grows, so do the demands on those who design, build and operate it.

From compliance and project planning to long-term optimization, success increasingly hinges on talent, partnerships and the ability to manage interconnected systems at scale.

Without the right people and processes in place, even well-funded initiatives risk stalling.

Cabling: the silent risk

One of the most overlooked barriers to AI-readiness is also one of the most foundational: cabling. 70% of operators believe poor-quality cabling will compromise their long-term AI infrastructure.

Cabling directly impacts performance, scalability and the ease of future upgrades. When done poorly, cabling can throttle workloads and introduce costly inefficiencies.

The risks are not just technical. Supply chain disruptions are affecting material availability, and 27% of operators cite a lack of skilled labor for cabling installation and maintenance.

Others are under pressure to align cabling with sustainability goals, with nearly a third (32%) focused on greener materials and solutions.

The problem isn’t just the execution, it’s the mindset. Cabling is often treated as a last-mile decision, rather than a strategic element of infrastructure planning.

The result? Shorter lifespans, performance bottlenecks, and expensive retrofits that could have been avoided with upfront investment.

A blueprint for holistic design

Meeting the demands of AI at scale won’t be achieved through siloed upgrades or short-term fixes. It requires a holistic approach to design that integrates every component of infrastructure from the start: power, cooling, cabling, site selection, operations, sustainability and staffing.

The consequence of neglecting this approach is already clear. 61% of operators have experienced costly design reworks due to fragmented decision-making. In contrast, 70% agree that holistic design will be essential to building AI-ready facilities.

The most forward-thinking operators are aligning around five core principles:

1. Define the purpose – Design around the data center's strategic business role, not just its technical requirements.
2. Create a shared vision – Align stakeholders across engineering, operations, finance and sustainability from the outset.
3. Design for adaptability – Build in modularity and flexibility to support evolving workloads and technologies.
4. Use the AI advantage – Leverage AI to optimize everything from energy use and performance to predictive maintenance.
5. Avoid false economies – Invest early in quality design and materials to avoid downstream inefficiencies and rework.

Data centers have stepped out of the background to become vital pillars of economic growth, national resilience and technological advancement.

The UK’s decision to designate data centers as Critical National Infrastructure – and its commitment to re-examining key planning decisions – signals just how central they are to the country’s digital future. Backed by £6.3 billion in global investment, this momentum is undeniable.

But the pressure to deliver is intensifying. Speed, cost control, sustainability and compliance are all in play. In this environment, those who embrace a holistic, forward-looking mindset will be the ones to lead.

They’ll deploy infrastructure that is scalable, resilient and future ready, not just built for AI today, but capable of adapting to whatever comes next.

We've featured the best AI website builder.

This article was produced as part of TechRadarPro's Expert Insights channel where we feature the best and brightest minds in the technology industry today. The views expressed here are those of the author and are not necessarily those of TechRadarPro or Future plc. If you are interested in contributing find out more here: https://www.techradar.com/news/submit-your-story-to-techradar-pro