Skip to content

Powering AI: Challenges and Opportunities for the UK Energy System

Thu 8th October 2026 | AI

Powering AI: Data Centres & the UK Energy System

The rapid growth of AI is bringing the relationship between digital infrastructure and energy firmly into focus. At our latest London event, the Scottish Energy Forum brought together perspectives from across the energy, infrastructure, advisory and real estate sectors to explore what growing data centre demand could mean for the UK energy system, and how the energy system might need to evolve in response.

The discussion began by establishing energy as the foundation on which the wider AI ecosystem depends. From computing infrastructure and data centres through to AI models and applications, each layer ultimately relies on access to sufficient, reliable power.

As AI becomes increasingly important to economic competitiveness, national resilience and strategic autonomy, access to the energy needed to support it is becoming an increasingly important consideration for governments and markets globally.

From grid constraints to new approaches to power

One of the clearest themes of the evening was the challenge of connecting new data centres to the UK electricity grid.

Long connection timelines, uncertainty around connection dates and the scale of demand already in the queue are creating challenges for both data centre developers and the energy system. There was recognition that reform is needed to distinguish between projects genuinely capable of progressing and more speculative capacity in the queue, while striking a balance between improving the connections process and creating additional barriers to investment.

Proposed reservation or commitment fees were discussed in this context. While stronger financial commitments could help prioritise credible, well-funded developments, concern was raised that significant upfront costs could favour the largest hyperscalers and risk making the UK less attractive to other developers.

These constraints are also changing how data centres think about power. Rather than relying solely on a conventional grid connection, developers are increasingly considering self-generation, behind-the-meter generation, storage and other flexible solutions.

The US provided an interesting comparison. Planned AI data centre capacity is increasingly clustering around areas with abundant gas resources, with speed to power and the ability to bring generation alongside new demand becoming important considerations. The UK faces a different set of circumstances, but the underlying question is similar: whether future data centres should increasingly be located where energy is readily available.

The role of gas and flexible generation

Gas was considered as one potential part of the solution, particularly where firm and flexible generation is required.

Different AI workloads can have different demand profiles, meaning that data centres do not necessarily represent a completely flat, continuous load. Alongside renewable generation, flexible sources of power may therefore have a role in responding to changing demand.

Rather than one technology emerging as the answer, the discussion pointed towards a range of possible configurations. These could include grid-connected data centres supported by storage or flexible generation, self-generation, or combinations of technologies designed to reduce permanent demand on the grid.

Gas was also discussed as a potential bridge while developers wait for grid connections or longer-term alternatives to become available. Fuel cells were highlighted during the Q&A as another technology attracting interest, including their potential to form part of a wider stack of technologies rather than operating as a standalone solution.

Could nuclear become part of the solution?

Small modular reactors were another significant area of discussion.

The potential attraction for data centres is the prospect of firm, low-carbon power capable of supporting continuous demand. Interest is already developing around the relationship between nuclear generation and data centres, although regulatory, planning and licensing challenges remain.

In particular, the discussion considered whether the current framework is sufficiently adapted to smaller nuclear technologies, rather than applying requirements developed for much larger conventional nuclear projects.

While SMRs could therefore become part of the longer-term energy mix for data centres, development timescales mean that other solutions may still be required in the interim.

Taking data centres to the energy

Location emerged as a recurring theme throughout the evening.

Historically, data centres have clustered around major demand centres and established availability zones. The growth of AI could begin to change that relationship.

Not all AI workloads have the same sensitivity to latency. Applications requiring extremely rapid responses may still need to be located close to users, while other workloads can potentially be located further away. This creates an opportunity for some data centre development to move closer to available energy rather than requiring increasing volumes of electricity to be transported towards traditional demand centres.

Scotland was discussed extensively in this context because of its significant renewable energy resources and existing energy infrastructure. Locating appropriate data centre workloads closer to generation could offer one way of making greater use of available power, particularly where latency is less critical.

However, access to energy alone does not determine where projects are built. Power prices, planning, community acceptance, occupier demand and the availability of commercially viable sites all remain important considerations. It was also noted that the UK’s relatively high power prices make some forms of AI activity, particularly large-scale training, more challenging to attract.

The discussion also highlighted the potential of former power station and other brownfield energy sites. Existing grid infrastructure and industrial land uses can make these locations attractive for new data centre development, while some sites may also offer access to other useful infrastructure.

Can data centres help the energy system?

While much of the debate around data centres focuses on the additional pressure they could place on the energy system, the discussion also considered what they might contribute to it.

One opportunity is greater demand flexibility. Large operators with data centre capacity across multiple locations could potentially route some computing activity between sites, while certain workloads could be scheduled at different times to respond to conditions on the grid.

Locating flexible demand in areas where renewable generation is constrained could also help make use of electricity that might otherwise be curtailed.

The addition of significant new electricity demand could have wider implications for system costs. One argument explored during the discussion was that increasing the electricity user base as the economy electrifies could spread system costs across greater overall demand.

Approaches in other markets were also considered, including policies linking large new energy users with additional renewable or flexible generation. Behind-the-meter renewable generation, gas generation and fuel cells are among the options currently being considered as part of the wider power mix for data centre developments.

Beyond electricity: heat and water

The discussion also considered the wider resource requirements and potential benefits associated with data centres.

Data centres produce significant quantities of heat, creating an opportunity for this to be captured and used by surrounding homes or businesses. The challenge is not simply making heat available, but establishing the infrastructure and commercial arrangements needed to transport it to potential users.

On large, master-planned developments where different uses can be designed together, integrating heat networks may be more straightforward. Connecting a standalone data centre to third-party users is more complicated, particularly when responsibility moves beyond the boundary of the data centre itself.

There was nevertheless a view that planning requirements, regulation and incentives could help encourage greater use of waste heat where appropriate.

Water use was also raised during the Q&A. While this could become a constraint for some developments, existing energy infrastructure may present opportunities. Some established energy sites already have water abstraction licences and associated infrastructure, potentially allowing existing resources to be reused rather than adding entirely new demand.

Planning, communities and public perception

Community acceptance was identified as another important challenge, particularly as the scale of proposed developments increases.

Large data centres can represent a significant change in land use, particularly in areas without an established industrial landscape. Former power station and other industrial sites may therefore have advantages beyond their existing infrastructure, with surrounding communities already accustomed to large-scale energy or industrial activity.

An audience question also prompted discussion about the language used to describe new developments. While AI is driving much of the current increase in data centre demand, data centres already underpin a much wider range of services used every day.

The discussion highlighted the importance of improving public understanding of what data centres do, why different types of data centre are required and what their development could mean for local communities. The experience of other forms of energy infrastructure was raised as an example of how public perceptions can evolve over time.

Who pays for the infrastructure?

The cost of supporting growing data centre demand was another key question raised by the audience.

New grid capacity, reinforcement and generation all require significant investment. The discussion considered how these costs should be divided between developers, users and the wider electricity system, with concern that new data centre demand should not have a detrimental impact on existing energy consumers.

At the same time, the commercial costs of developing and operating data centres ultimately form part of the economics of the digital services and AI applications they support. How those costs flow through the value chain will therefore remain an important consideration as the market develops.

Resilience, security and the case for UK capacity

The Q&A also explored the relationship between data centres, resilience and security.

For some AI applications, the need to keep sensitive or proprietary data within a particular jurisdiction could strengthen the case for data centre capacity in the UK and Europe, even where other markets offer lower energy costs.

This added another dimension to the discussion around location. Decisions about where data centres are built will not be driven by energy alone, but by the interaction between cost, infrastructure, latency, security, resilience and the requirements of individual workloads.

Balancing growth, affordability and decarbonisation

The discussion concluded by returning to the wider challenge facing the UK: how to accommodate potentially significant growth in data centre demand while balancing energy security, affordability, economic competitiveness and decarbonisation.

Rather than trying to determine a fixed amount of data centre capacity that the UK should accommodate, the discussion favoured creating appropriate guardrails within which the market can develop. Given how rapidly AI technologies, computing requirements and data centre designs are changing, attempting to set a fixed requirement today could quickly become outdated.

The interaction with the UK’s net zero ambitions was also explored. Growing data centre demand could place additional pressure on the energy system, creating a need to consider how new demand can be supported alongside low-carbon and renewable generation. At the same time, the discussion reflected the increasing importance of energy security, affordability and industrial competitiveness alongside decarbonisation.

What emerged throughout the evening was that there is unlikely to be a single solution to powering the next generation of data centres. Grid investment, renewables, flexible generation, storage, gas, nuclear and new approaches to locating and operating data centres could all have a role to play.

As AI continues to develop, the relationship between energy and digital infrastructure will become increasingly interconnected. The challenge for the UK will be to create an energy and infrastructure environment capable of responding to that growth, while balancing the needs of consumers, communities, the wider energy system and the UK’s economic ambitions.

Meet The Panel

Ronan Lambe

Partner & Event Chair – Pinsent Masons

Ronan advises on energy and infrastructure projects, with particular expertise in renewables, power procurement, grid connections and data centre developments.

Tong Wu

Advisor – Lambert Energy Advisors

Tong advises on M&A and strategic mandates across the energy sector, with a particular focus on AI infrastructure, power demand and hyperscaler energy procurement.

Sarah Stewart

Partner – Energy & Data Centres – Newmark

Sarah is a Chartered Surveyor with over 15 years’ experience advising on energy infrastructure, data centres, power generation and the wider energy transition.

Ian Cook

Head of Business Development – SSE

Ian has over 14 years’ international experience developing major energy infrastructure across power generation, renewables and energy storage.