Why the UK’s AI ambitions demand a new power paradigm
Author: Joe Peck
In this exclusive article for DCNN, Javier Cavada (pictured above), President & CEO EMEA at Mitsubishi Power, considers how private power networks and on-site generation could help data centre operators overcome grid constraints while supporting the UK’s AI ambitions:
Decoupling digital growth
In the UK, expanding digital and AI infrastructure is a strategic priority. The Government has indicated that the country will need at least 6GW of AI-capable data centre capacity by 2030 to support this ambition.
However, this digital growth is outpacing the capacity of the UK’s physical energy infrastructure, resulting in a growing mismatch between the country’s AI ambitions and the systems needed to support AI-powered data centres.
Electricity demand from data centres is already significant, at a time when the wider economy – indeed, society as a whole – is rapidly electrifying. Currently, data centres account for around 6% of UK electricity consumption, and the National Energy System Operator (NESO) projects that this could increase to 8.8% by 2030 as AI adoption accelerates.
At the same time, grid connection requests for demand-side projects have surged from around 41GW in late 2024 to 125GW by mid-2025, with approximately 50GW linked to data centre developments. This rapid build-up in the connection queue is creating significant congestion, with some large, high-capacity projects now facing delays of up to a decade.
All of this has prompted regulatory intervention, with NESO moving away from the previous “first come, first served” approach towards a more selective “first ready, first served” model for prioritising connection requests.
While this shift is welcome, it will take time to translate into tangible improvements on the ground. In the meantime, the key takeaway for developers is clear: securing a guaranteed power connection has become a far more significant constraint on new data centre development than access to land.
More broadly, the structural limitations of the UK’s centralised and ageing grid are emerging as a major barrier to delivering on the country’s AI and digital infrastructure ambitions.
The grid bottleneck and the competitiveness risk
The reality is that to ensure operational uptime from day one, operators can no longer rely solely on the UK’s national grid. Instead, delivering the power required to build and operate this critical infrastructure will increasingly depend on on-site energy parks and dedicated private-wire networks.
This challenge extends well beyond a single industry; it is fundamental to the UK’s ability to sustain a dynamic, modern economy. AI, cloud computing, and high-performance computing (HPC) have become core drivers of global competitiveness, and, in turn, access to reliable power is a decisive factor in where hyperscalers and technology firms choose to deploy capital.
These decisions shape long-term job creation and regional economic growth. Without sufficient power availability, the UK risks losing major digital investments – as well as the high-skilled employment they bring – to leading European markets.
Competitors in the Netherlands, Ireland, and the Nordics are gaining ground by offering faster access to power – a trend already evident in the Slough/M4 corridor, where connection moratoriums have pushed operators to look beyond traditional hubs simply to keep pace with demand.
One response is the deployment of high-efficiency gas turbine systems to help bridge this capacity and infrastructure gap. Gas turbines provide a practical interim solution, delivering reliable, dispatchable power at scale today while offering a pathway to lower-carbon operation as hydrogen and other low-carbon fuels mature.
Why private power models are becoming essential
By connecting dedicated power assets directly to the data campus via private-wire networks, operators can bypass multi-year utility queues, significantly compress construction timelines, and secure a predictable envelope of capacity. Operating behind the meter also provides a critical commercial advantage, shielding multi-million-pound infrastructure investments from volatile wholesale market prices and localised grid congestion, as well as enabling greater long-term cost certainty.
However, access to power alone does not fully resolve the challenge. AI workloads require continuous, 24/7 baseload stability – something intermittent renewables cannot deliver in isolation.
As the UK continues to scale wind and solar generation, managing intermittency becomes an increasing constraint, reinforcing the role of on-site gas turbines in providing immediate, dispatchable power to stabilise private networks.
Crucially, deploying on-site gas generation does not mean abandoning sustainability goals. Instead, it offers a pragmatic bridge to net zero. The industry is already shifting towards flexible thermal infrastructure that can meet current demand using natural gas, while remaining compatible with lower-carbon fuels.
Modern high-efficiency gas turbines, for example, can already operate on a 30% hydrogen blend, with engineered pathways to 50% and ultimately 100% hydrogen capability from around 2030 onwards, as technology and fuel supplies mature. This ensures that assets deployed today remain viable in a decarbonised future.
Embracing a new model of infrastructure self reliance
The UK cannot become a global AI leader if its data centres remain dependent on an increasingly constrained public grid. Colocated energy parks represent a shift from grid dependency to infrastructure self-reliance. By deploying on-site generation, operators can decouple build timelines from grid constraints while laying the foundations for a more resilient, future-ready, low-carbon digital economy.
In this context, digital sovereignty cannot wait for grid reinforcement. The sector is moving towards a model in which operators take greater control of their energy supply, ensuring both immediate resilience and long-term strategic flexibility.
For more from Mitsubishi, click here.
