AI: decarbonising data centres

Jon Healy at Salute explains why the UK’s AI superpower ambitions rely on decarbonising data centres

 

The UK Government’s ambition to establish itself as an AI leader is gathering momentum. Both the public and private sectors are investing billions of pounds in AI research and development, aiming to enhance innovation across various industries.

 

To get there, an important consideration is often overlooked: how does the Government plan to accelerate AI adoption without first ensuring that sustainable infrastructure is in place?

 

This is a very real, headline-driving challenge. It’s only recently that Amazon’s plans for a new UK data centre campus have raised fresh worries about grid capacity and sustainability, particularly when you consider that the facility could use as much power as the entire town of Burnley.

 

A study by the LSE and the Grantham Institute suggests AI could help cut up to 5.4 billion tonnes of carbon dioxide equivalent by 2035. This contrast highlights the crux of the problem: how to meet the growing demand for computing power while adhering to climate commitments. It boils down to the data centre industry and its partners to tackle the challenge of decarbonisation to turn these goals into reality.

 

The UK has been vocal about its AI ambitions, but it is far from alone. While facing many of the same infrastructural and environmental issues, governments throughout the EMEA region are rushing to expand their sovereign AI capabilities. From Germany to the UAE, there is a common push to build next-generation data centres, strengthen grid capacity, and meet tightening environmental regulations, all within energy markets under increasing strain.

 

 

Shared responsibility across sectors

Pressure on policy is growing throughout the region. Regulations like the Corporate Sustainability Reporting Directive (CSRD) in the EU are increasing the bar for standardisation and transparency in emissions reporting.

 

At the same time, operators find it more difficult to meet investment standards without deeper decarbonisation initiatives owing to the EU Taxonomy’s stringent energy-efficiency guidelines on what constitutes sustainable infrastructure.

 

Data centre decarbonisation is not limited to sustainability specialists or facilities teams. Operations, engineering, compliance and policy should all work in conjunction on this. From choosing the site and building it to sourcing power and maintaining equipment, each of these areas aids in reducing emissions over the course of a site’s lifecycle. Until a collaborative approach is achieved, we’ll keep seeing fragmented progress.

 

We also need to acknowledge that simply being more efficient is no longer enough. As AI training and demand grow, so will the need for energy. Operators need to adopt a whole-systems approach, focusing on renewable energy, designs that reuse materials, and clear tracking of all types of emissions.

 

 

Infrastructure and energy constraints

There are already bottlenecks in the infrastructure required to support AI at scale. Grid capacity is constrained in many high-demand regions, including the South East and London. According to some providers, new connections won’t be accessible for another ten years. As a result, the UK’s potential to boost local computing capacity is restricted.

 

Delays in planning and uncertainty about long-term energy availability hamper the development of new data centres. The UK runs the risk of turning into a net importer of computing power if these problems continue, thereby losing control and innovation to nations with more developed infrastructure plans. The UK cannot afford this situation if it hopes to maintain its global position.

 

The UK is not alone; in the Netherlands and Ireland, local resistance, environmental concerns and limited grid headroom have resulted in moratoriums or delayed permissions for new data centre builds. Without cross-border coordination and long-term energy investment, EMEA’s AI ambitions could run into the same constraints already evident in the UK.

 

 

Aligning ambition with sustainable delivery

While funding for AI research is essential, data centre capacity and sustainability investments must also be treated with equal importance. The incentives of embracing clean energy, renewing old infrastructure, and creating quicker grid connections must also be considered by policymakers. Clear planning frameworks can help construct new and effective facilities in underserved areas.

 

Aligning the growth of AI infrastructure with the UK’s climate commitments is also necessary. There is a chance that advancements in one area could jeopardise commitments in another if there is no clear guidance on how compute growth fits into national carbon budgets.

 

To stay competitive and climate-conscious, EMEA governments must align AI growth targets with realistic infrastructure planning. This requires strategic coordination between regulators, operators and power providers - not just within borders, but across them.

 

 

Managing complexity and emissions in parallel

Data centre operators are currently facing a two-fold problem: dealing with complex infrastructure while also trying to reduce emissions. Many operators are looking into options like modular construction, advanced cooling technologies, and the use of renewable energy sources.

 

But constraints from outdated decisions about existing infrastructure and inconsistent reporting standards across the industry often slow down progress. Data centres need to be more transparent and have stringent methods for measuring emissions during both the construction and operational phases.

 

To address these problems, setting benchmarks, writing detailed reports and sharing information across the industry are the need of the hour

 

 

A moment of opportunity

The UK data centre industry is at a critical crossroads. As both artificial intelligence and the need to reduce carbon emissions grow rapidly, there’s a significant opportunity for greater collaboration and innovation. The UK Government should make data centre infrastructure a national strategic priority. This will require a full, whole-systems approach that encourages close cooperation among the government, businesses, and energy providers.

 

The stakes are clear: to promote innovation, maintain our position in the global AI revolution and exercise environmental leadership, we must act now to be both scalable and sustainable.

 

Looking ahead, regional coordination across EMEA could make the difference between fragmented progress and strategic leadership. This is a moment of shared opportunity and responsibility to ensure AI transformation does not come at the cost of environmental or infrastructure collapse.

 

---

 

Jon Healy is Managing Director, EMEA at Salute

 

Main image courtesy of iStockPhoto.com and yucelyilmaz