Worldwide demand for digital infrastructure continues to surge, predominantly driven by the AI boom. However, data centre builders looking to capitalise on this generational hunger for greater compute capacity will enter 2026 facing myriad challenges, new and old. AI data centres are consuming extraordinary amounts of power, and AI workloads are starting to reshape the digital infrastructure that hosts them. Increasingly, these trends are coming into conflict with constrained power grids, supply chain woes, and stricter ESG requirements. The upshot is that data centres face unprecedented pressure on their power infrastructure.
In 2026, the industry will have to navigate this convergence of grid limitations, supply chain constraints, and evolving regulatory environments. Highlighting this rapidly evolving and uncertain landscape, we’re going to take a look at the five biggest power-related challenges shaping the data centre landscape, and the emerging strategies organisations are using to stay ahead.
1. T1 Market Power Constraints
Goldman Sachs Research forecasts global power demand from data centers will increase 50% by 2027 and by as much as 165% by the end of the decade, reflecting the massive worldwide surge in compute capacity required to support the booming AI sector. The problem is that the AI, data centre, and energy industry all move at different speeds.
Data centres can typically be built in one to two years. That’s still slower than AI companies would like, but the real challenge is that the electricity infrastructure needed to support new data centres takes much longer to get connected. The timelines required to expand transmission networks, substations, and generation capacity stretch far further into the future. In the EU, wait times for securing a grid connection can range from two to ten years. More and more, this mismatch is becoming a critical bottleneck. Meanwhile, hyperscale and AI-focused developers continue to cluster near major cities where demand is highest, placing additional pressure on already stressed local grids.
Racks supporting AI training workloads can exceed 40 kW today, trending toward 85 kW, with 200–250 kW per rack projected by 2030, and long-term visions discussed by Google exceeding 1 MW racks. These power densities are running up against hard physical and regulatory limits. Several major hubs, including Dublin and Amsterdam, have already paused new grid connections due to capacity constraints before the AI boom. Increasingly saturated markets in the FLAP-D region are likely to follow suit.
Put simply, Tier 1 markets can no longer support growth at the scale or speed required by AI infrastructure. Developers can’t continue building in the same locations in the same ways as before.
In 2026 and beyond, data centre companies need to look seriously at new geographic hubs with direct access to available power. Regions such as the Northern Ireland coast could offer proximity to generation assets and more favourable grid availability. This macro-level strategy helps decongest Tier 1 markets and unlock faster development cycles.
2. AI Training Is Redefining Data-Centre Power Consumption
It’s no secret that AI increases the overall energy demand of data centres. However, less well known is the fact it also reshapes how and where power is consumed. AI training workloads require enormous, sustained electricity use but are (a bit) less latency-reliant than cloud or AI inference workloads, enabling facilities to be placed far from population centres. As a result, there’s a new generation of data centres dedicated especially to training large AI models that, while physically larger and more power-intensive, can be located in more remote areas. Outside congested, largely urban, hubs, data centres can capture advantages like lower land costs, access to water, and abundant renewable power.
However, building any data centre (let alone one big enough to support mass AI training and similar applications) anywhere off the beaten track presents logistical issues. Doing so creates complexities around construction, manufacturing logistics, and commissioning, and traditional approaches to building in Tier 1 markets often fall short.
Data centres must find ways to deliver massive compute capacity in remote locations while overcoming logistical challenges in construction and deployment. AI-focused sites require a different design and build methodology. Companies that bring modular construction expertise and experience delivering remote, large-scale facilities can help accelerate deployment, reduce risk, and simplify the thorny prospect of building far outside established industrial hubs.
3. Land Scarcity Demands Greater Flexibility
Tied into the first two challenges is the problem of land. Data centres are big business, but more prosaically, they’re very big buildings, often with attendant outbuildings for hosting UPS and backup generators. Securing suitable plots for data centre development is becoming increasingly difficult and expensive, especially in or near major metro areas. Sites with the right combination of access to power, fibre, zoning, and environmental compliance are increasingly scarce, driving up costs across multiple markets.
Developers are finding themselves faced with interminable permitting cycles, not to mention growing community scrutiny around noise, water usage, and sustainability.
These land constraints often force operators to make trade-offs: building more capacity in multiple locations with smaller footprints, compromising on ideal site layouts, or locating closer to power generation at the expense of also being close to an increasingly hostile general public.
Flexibility is becoming a key differentiator. Leveraging modular, highly customisable power systems makes it easier to adapt infrastructure to unconventional or constrained sites. This helps reduce the burden of site selection and enables operators to maximise usable capacity within limited or irregular parcels of land.
4. Tightening Cross-Border Regulations and Supply Chain Pressures
The global data-centre supply chain is already under significant strain due to surge in demand. Now, data centre companies (often international firms with sprawling supply chains) are having to contend with heightening cross-border complexity. Continued post-Brexit fallout is snarling up the border between the UK and Europe, and US tariffs — bad enough on their own — have ushered in a new era of frosty trade relations and more isolationist policies. Setting aside the increased cost of moving construction materials and data centre components around the world, customs complexities are also driving up lead times. Critical components such as switchgear, transformers, batteries, and high-density compute hardware all face growing bottlenecks. Deloitte reported recently that large original equipment manufacturers are increasingly locking themselves into multi-year agreements to produce key components, like transformers, switchgear, power management equipment, and power generation systems with single clients. As a result, these critical components are sold out, with restocks not expected to arrive for several years.
All of these challenges mean steep increases in lead times across core electrical infrastructure. These delays directly threaten deployment schedules for new builds and expansions.
To navigate this environment, data centres are leaning on partners with localised manufacturing, diversified supply networks, and the ability to operate across regulatory borders. Companies with operations in Northern Ireland, for example, enable clients to take advantage of UK-based trade agreements (including with the US) as well as Northern Ireland’s unique relationship with the EU offering operators a way to mitigate cross border complexity Partnering with companies that have in-house engineering and assembly capabilities also reduces reliance on constrained international supply lines.
5. Evolving ESG Regulations Increase Operational and Design Pressures
Governments are accelerating ESG mandates as net-zero deadlines approach, with the EU typically leading the way. For data centres, this means tighter scrutiny on power efficiency, emissions, water usage, generator runtime, and backup-power resilience.
At the same time, operators must secure access to a growing but still limited pool of clean energy sources. Variability in renewable generation introduces added complexity for UPS systems and backup infrastructure, creating concerns about reliability under fluctuating loads, especially when regulatory frameworks restrict generator size and design specifications.
Adopting modular, easily serviceable power systems allow for rapid replacement and minimal downtime. Predictive monitoring capabilities enhance visibility across power-distribution components, improving both “time to detect” and “time to resolve.”
Flexibility and Resilience in the Face of a Changing Power Landscape
In 2026, the power landscape for data centres will be defined by scarcity, complexity, and accelerating demand from AI. Operators that explore new locations, embrace modularity, rethink supply chains, and keep a watchful eye on evolving ESG requirements will be best positioned to deliver reliable capacity at scale.
Partnering with organisations that have expertise related to remote builds, modular power systems, and cross-border operations is becoming an important competitive edge for data centre companies looking to navigate the challenges to come.
