Securing enough power is only part of the challenge facing data centre developers. Jamie McAinsh, Founder Director at Aurora Utilities, explains why certainty over when and how that power will be delivered has become just as important.
Looking back 10 years, the majority of UK data centres were clustered in and around London. Fast forward to 2026, and power availability – and the certainty of securing access to it on time – has overtaken location as a priority for operators, as AI reshapes where and how new data centres are built and connected.
While the power-hungry nature of these operations has been widely reported, the complexity of securing the large volumes of power needed to run them is less well understood. Behind the scenes, it is creating serious challenges for operators.
A survey of leading UK data centre developers, investors and operators published last year found that 74% believe national infrastructure delays are curtailing projects, while 50% reported changing project locations because of connectivity and cost constraints.
These statistics point to a common reality: securing the right grid capacity at the right time is now one of the most significant challenges developers face.
The evolving priorities of data centre operators, together with regulatory reform on the horizon, are changing the UK connections landscape. In response, new approaches to securing and delivering power are emerging. Five developments illustrate this shift.
1. The new geography of data centres
Proximity to customers, fibre infrastructure and low-latency requirements previously dictated site selection. This meant that data centres were predominantly based in and around London, including to the east in Docklands and Essex, the north in Oxfordshire and Hertfordshire, the west in Slough and Berkshire, and the south in Croydon.
As AI training, machine learning and high-performance computing environments can allow for greater geographical flexibility, location may become less important for some workloads.
Developers are therefore looking at sites through a different lens: where can they secure the grid capacity required for an energy-intensive operation, and how can this be achieved sustainably over time?
In 2026 and beyond, data centre development is becoming more geographically dispersed. Scotland, for example, has up to 24 hyperscale sites planned. Its onshore wind resources make it an attractive destination for data centre investment. By consuming power closer to where it is generated, new facilities may be able to make more efficient use of renewable energy, reduce curtailment and help balance the electricity system.
2. Capacity is only half the story: navigating land rights
While network capacity constraints are widely discussed in the data centre industry, there are other practical considerations involved in securing power and delivering it to a site. Power on paper can look very different from power delivered in practice, particularly when multiple stakeholders are involved.
Securing a connection offer is only one part of the process. The time it takes to connect a site is often underestimated. It may require cabling across public highways, private land, transport corridors or multiple local authority boundaries, each bringing additional permissions and legal agreements that can take years to secure.
Obtaining rights for cabling across multiple landownership boundaries can add months to project timelines. Developers therefore need to work with network partners that can identify potential delays early and recommend alternative routes before those delays materialise, including options that may not initially be obvious.
Experience of this process and advance planning are integral to project success and, in many cases, will determine when a new facility can begin generating revenue.
3. Getting it wrong can be costly
Network operators commonly introduce milestone-based arrangements and ramping agreements to manage constrained capacity, so missed milestones can create uncertainty.
If deadlines are repeatedly missed, an operator may find that its allocated power is withdrawn, meaning it would need to apply again. This is significant because costs can escalate rapidly once project timelines begin to slip.
One industry report estimated that every month of delay on a typical 60 MW data centre project can result in up to £10–11 million in lost revenue, cost overruns and contractual penalties. Scaled to a 250 MW hyperscale facility, the cost of a prolonged delay could run into hundreds of millions of pounds. A 12-month delay could potentially exceed £1 billion once financing costs, deferred revenues and programme overruns are taken into account.
The wider economic impact is also becoming increasingly evident. A recent International Energy Agency report found that approximately 20% of planned data centre projects globally face significant delays because interconnection queues can extend to a decade in some regions.
4. Reform is disrupting the market
At the end of last year, we conducted a survey of UK infrastructure decision-makers, from investors to developers, and found that 62% believe AI-driven electricity demand will outpace grid updates by 2030.
It is hardly surprising, then, that pressure on the Government and regulators is intensifying as data centre operators call for grid connection reform.
In June 2026, Ofgem published a proposal for a new Independent Transmission Owner (ITO) licence.
While an ITO licence would not create additional grid capacity, it could open the market to independent regulated organisations able to accelerate the delivery of connection infrastructure once capacity has been allocated.
The UK has addressed similar bottlenecks before. In the 2000s, housebuilders argued that much-needed new homes could not be connected quickly enough because traditional distribution networks were facing substantial connection backlogs.
The Government subsequently introduced the Independent Distribution Network Operator (IDNO) licence, changing how new developments connect to the grid. It is estimated that between 80% and 90% of new developments are now connected by IDNOs in the UK.
Today, AI and data centre growth are being constrained by transmission bottlenecks, but the sector may be approaching a similar shift if an ITO licence is granted. This is particularly relevant to hyperscalers, whose power requirements increasingly extend beyond distribution-level connections.
5. A growing need for flexibility
For major data centre operators choosing a network partner, cost is not always the primary consideration. Developers I have spoken with point to three main priorities: the delivery track record of the network operator’s team, confidence in connection milestones and commitments, and the ability to accommodate a phased connection.
Mechanisms such as Asset Value Payments (AVPs), available through an IDNO partnership, can also affect project economics by returning value to customers when network assets are adopted. Depending on the scale and circumstances of a project, these payments can amount to several million pounds.
For many large-scale data centre developers, however, such financial considerations are secondary. What matters most is confidence that power will be delivered when promised and in line with commitments made to customers and investors.
A power strategy, not just a power connection
Ultimately, the UK’s ability to capitalise on the AI opportunity will depend on far more than servers and software.
Power strategy is now being considered much earlier in the development process. Capacity remains critical, but certainty is becoming an equally important factor in determining whether projects move forward on schedule or grind to a halt.
The projects that succeed over the coming decade will not necessarily be those with the largest campuses or the most advanced technology. They are more likely to be those underpinned by realistic connection timescales, early engagement with relevant stakeholders and a power strategy embedded in the development plan from day one.
