AI boom risks running Britain dry
AI is central to the UK’s growth plan – but that ambition depends on a data centre boom. While most attention has focused on their huge energy demands, Jon Chappell, deputy director at Water UK, says a bigger risk is being overlooked: water.
The government has made artificial intelligence central to its growth plan, including a trebling of the country’s data centre capacity by 2030. It has classified data centres as Critical National Infrastructure and set up AI Growth Zones, with more than 200 applications from local authorities. There is real ambition here, and water companies want to help deliver it.
But there is a problem that appears to be going unnoticed. Data centres need water – often very large volumes of it – to cool their systems. And right now, the country has no plan for where that water will come from.
Look at the government’s own strategies. Its policy for AI Growth Zones is dedicated to accelerating grid connections and bringing down energy prices. It does not mention water once. Neither does the AI for Science Strategy. Nor the AI Opportunities Action Plan. Three flagship documents, all silent on a resource without which data centres cannot run.
Jon Chappell, deputy director, Water UK
Jon Chappell, deputy director, Water UK
The government has made artificial intelligence central to its growth plan, including a trebling of the country’s data centre capacity by 2030. It has classified data centres as Critical National Infrastructure and set up AI Growth Zones, with more than 200 applications from local authorities. There is real ambition here, and water companies want to help deliver it.
But there is a problem that appears to be going unnoticed. Data centres need water – often very large volumes of it – to cool their systems. And right now, the country has no plan for where that water will come from.
Look at the government’s own strategies. Its policy for AI Growth Zones is dedicated to accelerating grid connections and bringing down energy prices. It does not mention water once. Neither does the AI for Science Strategy. Nor the AI Opportunities Action Plan. Three flagship documents, all silent on a resource without which data centres cannot run.
More than three quarters of the UK’s data centre capacity sits in the water-stressed east and south east of England, and the potential water demand from data centres is significant. If the government achieves its ambition to treble data centre capacity by 2030, peak demand from data centres alone could reach 77 million litres a day – almost four times the daily output of the new reservoir under construction at Havant Thicket. Even on current average efficiency, that figure is 42 million litres a day, twice Havant Thicket’s output.
The timing makes it worse. Data centres need water most on hot summer days, when they cannot rely on cooler outside air. But those are exactly the days when every other user needs more water too. If peak demand rises without a matching increase in storage and treatment capacity, water companies could struggle to meet demand on the hottest days of the year.
So why can’t water companies simply build for this demand they can already see coming? Because the system won’t let them. The document that assesses future water needs – the Environment Agency’s National Framework for Water Resources – explicitly excludes data centres from its forecasts. Water companies are required by law to plan investment in water resources, such as new reservoirs, within those forecasts, so they cannot invest ahead of demand the official framework does not acknowledge.
Even though the population and economy are growing, water companies are forced to plan on the assumption that the amount of water businesses need will fall by 9% by 2038 and 15% by 2050, even as the government bets the economy on water-hungry data centres.
This is why our evidence to the Environmental Audit Committee warns that new data centre growth in water-stressed areas may be impossible without policy change, and that the country’s need for data centres will almost certainly be thwarted without changing the rules.
The good news is that this is entirely fixable if government acts now. Three changes would make the difference.
First, reform the forecasting framework so that data centres are properly planned for, rather than excluded and ignored. Second, introduce water efficiency standards and demand greater transparency from data centres in water-stressed areas. Third, reform developer charging so that data centres pay the full network costs they impose, rather than those costs falling on household bills.
Household bills are already rising to fund a record investment programme. Under the current rules we are all subsidising the cost of connecting new data centres to the network too. So unless they are made to pay for the supply they need, the bill for powering the AI revolution will arrive at everyone’s door.
Water is not a barrier to growth, but it is a prerequisite for it. Water companies are investing to secure our water supplies and support economic growth. We want to power Britain’s AI revolution too, but we can’t until the government plays its part.
Britain cannot build the digital economy of the future on forecasts that pretend this huge new water user does not exist. The sooner we fix that, the sooner the country can get on with the growth it needs.
