Energy moonshot needed for UK to become AI superpower

The government’s new ambition for the UK to become an AI superpower is a compelling vision to unlock growth and improve productivity.

Yet, in a world where a single AI prompt requires nearly ten times the energy as a Google search, the AI Opportunities Action Plan has thrown the ongoing question of energy capacity and grid connections into the spotlight.

High levels of development in AI and off-site computation are expected to increase data centre electricity demand by fourfold from today to 2030, according to the National Energy System Operator.  But meeting the UK’s 2030 clean power target is already at the edges of feasibility. The government is right to convene an AI Energy Council, which must plug in to the spatial energy planning already underway by NESO: powering the AI revolution needs a long-term plan to realise its potential for the UK economy within a wider strategy for a clean, secure and affordable energy supply for 2030 and beyond.

Bottlenecks to Big Data success

To briefly reiterate the scale of the UK’s net zero energy transition: we must construct five times more grid infrastructure in the next six years than has been built in the previous decade; and balancing a more complex system means a major ramp-up in flexible low-carbon generation and energy storage capacity, alongside significant changes in demand response and unprecedented public support.

Scaling up AI data centre development will underpin the UK’s ability to establish itself as a global player, yet locating and connecting data centres to the grid has already been a notable challenge for some time.

Planning delays and the prioritisation of connections remain two fundamental factors to delivering the capacity needed for economic growth and sustainable development across the UK. The forthcoming Industrial Strategy and proposed planning reforms will be essential elements in overcoming these critical bottlenecks that hinder both net zero targets and AI ambitions for both data centres and the critical energy infrastructure to power them.

Alongside proposed planning reforms to prevent lengthy project delays, in many parts of the UK the queues to connect industry, housing and infrastructure onto the grid stretch out into the 2030s. Revising queue management arrangements for grid connections will also ensure that the right critical infrastructure and strategic assets are hooked up to the grid faster.

A nuclear resurgence

Nuclear has rightly been identified as a vital technology to fuel the surge in AI sustainably and meet the 99.999 per cent availability demands of data centres. The scale and agility of Small Modular Reactors (SMRs) have positioned them as a central solution, and investment from big tech is already positively reshaping the pace of commercialisation of new technologies across the world. For example, AtkinsRéalis is working with public energy agency Energy Northwest to deliver 12 SMRs for Amazon and X-energy in Washington, US.

While new energy technologies develop at pace, the push for big data requires more power in far shorter timescales. This presents potentially exciting new business cases to extend and repurpose the life of operational power stations to create capacity for data centres in the interim, until SMRs are selected and permitted in the UK. Looking across the Atlantic again, the Three Mile Island nuclear power plant is preparing to reopen following a 20-year deal with Microsoft, a solution which could be viable in the UK too, attracting new investment and extending assets to bridge the gap.

Whether providing new capacity or new life to existing assets, the momentum to power the potential of AI must be used to accelerate the UK’s energy transition as a whole. Optimising the long-term success of such industrial developments cannot though be at the expense of the delicate balance of affordability and stability of our future supply. The criticality of energy infrastructure must be assessed against the needs of homes, businesses and system stability across the country. 

The AI moonshot

Government has traditionally approached new technologies with caution. Yet the momentum generated by its bold aims around AI could prove essential to delivering growth and opportunities across the UK. It’s a complex challenge, but coordinated action on planning, nuclear development and private investment can put the UK on a path to powering both its AI ambitions and a broader energy transition.

This article was originally published in The Engineer.