The UK’s AI industry is running hot right now. Established tech behemoths like Microsoft are choosing to set up shop in London, the UK’s best minds are at the helm of leading AI startups, and the sector is awash with VC cash.

Credit for this success must be paid in part to the government. The UK has stimulated investment with tax breaks, supported emerging talent with research subsidies, and avoided the heavy-handed innovation-dampening regulatory position of the EU. These all paved the way for the private sector’s dynamism and innovation to power the country into the top tier of AI hubs.

However, if you peek under the floorboards of this success, worrying signs of rot have begun to set in – which, if the government fails to address them quickly, risk bringing the whole structure down.

The AI sector’s demand for energy to feed the ever-increasing demand for compute, risks overwhelming the country’s ailing supply grid and brittle suppliers.

Fortunately, the new government is keenly aware both of the value of the AI sector to the UK’s economy, and also the risk that the power supply crunch poses to industry. I was recently invited to the government’s AI Opportunities Action Plan Roundtable at Number 10, and the initiative demonstrates this government’s commitment to supporting the private sector and building on the industry’s success.

However, a key player was missing from the table – the energy sector. Any discussion on the UK AI’s future has to involve representatives from the energy industry so that tech, energy, and the government can constructively work towards relieving the acute pressure on the power supply.

AI is considerably more energy-intensive than traditional computing. One ChatGPT search uses up to ten times more electricity than a Google search. This discrepancy is driving a surging demand for power supply, with Goldman Sachs estimating that data center power demand will grow 160 percent by 2030, making up 3-4 percent of global use.

The problem that the UK faces is that its energy grid is outdated, and simply can’t handle this level of power demand, which creates severe bottlenecks for compute. The average age for transformers in the UK is 63, and current estimates are that the grid needs £60bn worth of investment just to meet the UK’s de-carbonization targets for 2035, let alone its AI ambitions.

The government is too cash-strapped to meet these demands itself – so it must work arm-in-arm with private sector energy suppliers to ensure that supply meets demand, but also in a sustainable way that guarantees the long-term functioning of these data centers.

What looks like a crisis can also be an opportunity, especially for green energy suppliers, whether they are the sustainable arms of legacy energy multinationals or newer energy scaleups. They should be clamoring at the chance to supply power-hungry data centers. The market is huge, and the data center customers are cash-rich enough to pay for the transition.

The first step of this is obviously modernizing the grid, so that any increase in energy produced by the green energy suppliers can be routed to data centers. This will take some upfront cash, but both the green energy and data center industries should be prepared to support the government on this – there are huge upside returns available once the grid’s capacity is increased.

But private-public sector partnerships can go much further than that. The government can start to offer tax incentives for data centers that use a certain proportion of green energy and also for the same energy providers who choose to supply the data centers.

Additional tax incentives could be provided to deploy sustainable data center infrastructure, such as cold aisle containment and liquid cooling that reduces the amount of energy required to operate. This would stimulate long-term strategic growth of sustainable data center infrastructure.

Some paired-back regulations in terms of planning and connecting to the energy grid would also make this transition faster. Currently, buying land and then connecting it to the power supply can be onerous, although not to the extent of other countries. The process is particularly time-consuming for major projects, as it currently takes between 12 and 14 years for new transmission lines to go from conception to being switched on. If the UK could streamline this process further, it would be an even more attractive location for data center providers to build in.

These are some short-term solutions that will work to alleviate some of the pressure as soon as possible. Long-term though, if the UK wants to be a dominant player in the AI industry, the future of data centers has to be nuclear. This is complex, but there is no other source that can hope to meet the long-term demands of AI-fueled computing demand sustainably.

The UK has all the pieces in place to become an AI superpower. It has the talent, the finance, and the tech pedigree to do so. It just needs some targeted attention from the government to ensure that the energy supply can meet these weighty aims.

There’s no way of getting to the top without large amounts of energy consumption – which must be done sustainably. The government must include the energy sector in its plans and start to work constructively together to power the UK’s AI sector forward.