How data centres are taking energy generation into their own hands
9 June 2026
As AI-driven data centres place unprecedented demands on electricity networks, Javier Cavada, president & CEO EMEA at Mitsubishi Power, argues that the UK must embrace flexible, co-located power generation and hydrogen-ready technologies to deliver the reliable energy infrastructure needed for sustained digital and economic growth.
Artificial intelligence (AI) is reshaping the European energy landscape faster than many anticipated. As economies undergo structural electrification, demand for power is rising across the board and nowhere more sharply than in power-hungry data centres.
Over the past five years, global data centre electricity consumption has grown by around 12% annually. According to the International Energy Agency, demand is expected to double again by 2030 as AIspecific computing accelerates.
This surge is driven by the nature of AI workloads. Today’s AI racks in data centres can draw between 10 and 20 times more power than traditional enterprise racks, with systems rapidly moving toward megawattscale densities. More importantly, these workloads can fluctuate within seconds, creating sudden “burst” demands that place significant pressure on local electricity networks. Because AI facilities cannot tolerate downtime, cooling and support systems must also run continuously. The result is a combination of high intensity and unpredictability that the UK’s power infrastructure was not originally designed to handle.
The UK’s grid bottleneck
In the UK, this challenge is becoming increasingly visible. Grid connection queues have lengthened significantly, with some projects facing waits of up to a decade for access to sufficient capacity. In London — and across major data centre hubs including Frankfurt, Amsterdam, Paris, and Dublin —available capacity is already exhausted.
This raises a critical question: can the UK deliver firm, reliable power at the pace demanded by AI-driven growth?
Why data centres must look at other options
One option is to develop on-site generation adjacent to the data centre, or to create private-wire systems that connect dedicated generation directly to the facility. Increasingly, new data centre developments are being designed from the outset to integrate power generation, energy storage, and digital infrastructure.
These “behindthemeter” approaches help avoid years-long delays associated with grid connections, while allowing locally generated power to be delivered directly into a data centre’s private electrical network. There are already established precedents in sectors such as healthcare, defence, and other critical infrastructure, where organisations have reduced reliance on congested networks and accelerated project delivery.
Behind-the-meter systems can also reduce the 6% to 8% energy losses typically associated with long-distance transmission. While the UK and Europe have significantly expanded wind and solar capacity, these intermittent sources cannot reliably provide the constant, dispatchable power that AI-driven data centres require without prohibitively expensive battery storage.
The reality is that no single solution will meet the demands of AI infrastructure. What is required is a balanced mix: renewables for long-term sustainability, and flexible thermal generation to ensure stability and reliability.
In the near term, natural gas remains one of the most practical sources of firm, dispatchable capacity to keep data centre operations continuously online. Over time, hydrogen presents a credible pathway to deep decarbonisation without compromising reliability. This is not a choice between technologies—it is an ecosystem where each plays a complementary role.
Catalyst for a smarter grid
This shift in how data centres approach energy generation has implications far beyond the sector itself. When equipped with modern, hydrogenready gas turbines, colocated power plants can deliver the fast response and operational flexibility that a renewablesheavy grid requires.
Today, many of these turbines are capable of operating on blended fuels — typically up to around 30% hydrogen with natural gas — with a clear engineering pathway toward 100% hydrogen firing capability.
Such facilities can help stabilise local networks during periods of stress, support the integration of intermittent renewable generation, and in some cases export surplus power back to the grid. In this way, the same assets that enable AI workloads can also strengthen overall system resilience as electrification accelerates across transport, industry, and buildings.
How today’s choices shape tomorrow’s grid
The rise of AI is forcing a re-assessment of how electricity systems are planned and operated. Meeting this demand requires thinking differently about both scale and resilience.
By deploying co-located, flexible power solutions today, the UK’s data centre sector can help drive the evolution of more responsive, decentralised energy systems — better able to balance supply and demand in near real time. What matters now is execution at scale: building the infrastructure that will underpin both digital and wider economic growth for decades to come.
