Aggreko: Could Decentralised Energy Power AI?

The global competition in artificial intelligence is facing a power dilemma rather than technological limitations. As AI advances, the infrastructure supporting these innovations faces pressure to adapt alongside soaring demands for electricity.
In the digital economy, data centres serve as vital components. Their need to expand is not limited to computation and storage; energy demands have surged to keep them operational. This new reality presents considerable challenges across the industry.
Is energy the new tech bottleneck?
The demands of AI workloads are unlike those from traditional cloud or enterprise applications. They require high levels of power for model training and inference, with energy consumption often far surpassing that of typical IT environments.
In Europe, and particularly in the UK, the ability to secure grid connections quickly is proving problematic. Key figures in the industry suggest that without addressing these issues, substantial investments in AI and digital infrastructure could bypass the region altogether.
Digital Realty, a global leader in data centre development, is lobbying for a systemic overhaul of the UK's energy grid and accompanying planning processes.
Aggreko, a major player in decentralised power solutions, echoes this sentiment in its Bridging the Energy Gap for European Data Centres whitepaper. The risk of falling behind in AI infrastructure looms large without stable power solutions.
“Our conversations with data centre owners and operators show Digital Realty’s concerns are echoed throughout the sector,” says Billy Durie, Global Sector Head for Data Centres at Aggreko.
“As with a great many other places, AI is set to have a transformative impact on the data centre industry. But if it is to do so, reliable power sources are key.”
Embracing decentralised energy
Aggreko highlights decentralised energy as a viable solution for addressing immediate energy needs. This approach involves data centre operators using on-site or modular energy generation methods to enhance resilience and flexibility, rather than depending solely on the national grid.
Current technologies being adopted include:
- Stage V HVO-fuelled generators: These use cleaner fuels, thereby lowering emissions without sacrificing reliability.
- SCADA-based monitoring and controls: These systems offer remote management and tight integration into facility operations, crucial for maintaining efficiency.
- Temporary bridging power systems: These are essential during construction and commissioning phases when grid connections are still pending.
According to Aggreko, these methods are not merely interim solutions but represent a strategic repositioning of energy as a core technological component. Just as businesses employ hybrid cloud models to enhance adaptability and dependability, decentralised energy offers a flexible approach to powering data infrastructure.
AI's critical influence
The rapid expansion of AI necessitates a reevaluation of energy procurement and infrastructure development. Aggreko’s findings illustrate the urgent need for bridging power solutions due to elongating grid connection queues that thwart immediate demand.
“The lengthening queues for grid connections provide proof that these sources cannot be guaranteed via the grid,” Billy says.
“While in the long run planning reforms will make it easier to deliver the necessary infrastructure needed to power facilities handling AI-heavy workloads, bridging power is required in the short- to medium-term.
“It is essential to ensuring UK data centre construction can continue to meet skyrocketing demand and not miss out on the huge levels of investment that come with AI technologies.”
Optimising procurement strategies
Aggreko emphasises the importance of reassessing procurement strategies due to challenges in securing permanent decentralised energy infrastructure in key markets. With supply chain constraints affecting procurement, temporary equipment hire is proposed as a feasible short-term strategy.
“Data centre construction and commissioning must continue as unhindered as possible if facilities are to be brought online within strict project timescales,” Billy says. “However, grid connection delays and a scarcity of decentralised energy solutions available for outright purchase means deadlines may be missed.
“Yet power is power, whether it comes from the grid or a hired generator. It is therefore vital project stakeholders and facility operators investigate different strategies for procuring necessary energy supplies.
“Leveraging third-party expertise within the supply chains can help these organisations access the expertise and capacity to swiftly deploy energy-efficient and scalable technologies on a hire basis, so they can meet specific site needs.”



