Key Discussion Points and Main Takeaways

The panel, ‘Building Sustainable Infrastructure: Powering Green Leadership’ brought together Lorraine Mackin, Tom Harris, Allan Bosley and Roger Dickinson to examine the infrastructure needed to support Britain's AI ambitions. The discussion explored energy costs, grid capacity, sustainability, government procurement and the importance of building public confidence in AI infrastructure.

Members can watch the full video here

Addressing Infrastructure and Energy Constraints

High electricity costs and limited grid capacity were identified as significant barriers to expanding the UK's AI infrastructure.

These challenges affect the commercial viability of new data centres, potentially discouraging investment and limiting Britain's ability to compete internationally.

The discussion emphasised the need for greater co-ordination between government, energy providers, regulators and infrastructure developers to accelerate projects and ensure sufficient capacity to meet growing demand for AI technologies.

Measuring Efficiency and Sustainability

The discussion highlighted the need to measure AI efficiency according to successful outcomes rather than simply the amount of energy or computing power consumed.

Technical measures such as energy consumption per token do not necessarily reflect whether an AI system has delivered a useful result. Measuring energy expenditure against completed tasks and business outcomes could provide a more meaningful assessment of efficiency and sustainability.

Opportunities to improve data centre sustainability were also explored, including waste heat reuse and closer integration with energy infrastructure. Solutions should reflect local conditions and the specific requirements of individual developments.

Using Government Procurement to Stimulate Investment

Government purchasing power was identified as an important mechanism for accelerating AI infrastructure development.

As a major technology customer, particularly through the NHS and wider public sector, government could help create demand for AI services and provide greater certainty for infrastructure investment.

Aggregating demand across departments and local authorities could strengthen the commercial case for new developments and support applications for grid connections.

Greater collaboration between government, regulators and infrastructure providers could also help reduce delays and encourage long-term investment.

Building Public Understanding and Community Confidence

Clearer communication about the benefits and environmental impacts of AI infrastructure was considered essential to securing public support.

The discussion highlighted the importance of explaining how AI investment can improve healthcare, public services and productivity, rather than focusing exclusively on infrastructure constraints and potential risks.

Data centre operators were encouraged to communicate more effectively about energy and water efficiency, using accurate evidence that reflects UK operating conditions rather than potentially misleading international comparisons.

Early engagement with local communities was also emphasised, including discussions about environmental impacts, economic opportunities and appropriate community benefits.

Supporting Decentralisation and Sovereign AI

The increasing decentralisation of computing infrastructure was identified as a significant development for the UK's AI ecosystem.

While hyperscale providers have traditionally dominated the market, investment in sovereign cloud, neocloud and enterprise data centres is creating opportunities for a more distributed infrastructure model.

Combined with the growing availability of open-weight AI models, this could give organisations greater flexibility over where and how they deploy AI technologies.

The UK will need infrastructure capable of supporting this transition, allowing businesses to choose providers based on cost, availability, energy consumption and sustainability.

Conclusions

The discussion reinforced that Britain's AI competitiveness will depend on developing infrastructure that is commercially viable, environmentally sustainable and capable of supporting increasingly diverse AI technologies.

The principal conclusions were:

  1. Address high energy costs and grid capacity constraints to attract investment and support infrastructure growth.

  2. Measure AI efficiency through successful outcomes rather than computing consumption alone.

  3. Use government procurement and aggregated public sector demand to strengthen the investment case for new infrastructure.

  4. Improve public communication and local engagement to build confidence in data centre developments.

  5. Invest in distributed computing infrastructure to strengthen provider choice, operational resilience and sovereign AI capabilities.

For the UK, building competitive AI infrastructure will require a co-ordinated approach across government, industry and local communities. Aligning energy planning, public procurement and infrastructure investment with measurable economic and social outcomes will be essential to ensuring that AI development delivers lasting benefits for businesses, public services and the wider economy.