Loading summary
A
Millions of us have flocked to the cinema to see fantastical portrayals of civilizations fighting over water. Whether it's James Bond in Quantum of Solace or the battle to hoard moisture on the desert planet of Arrakis in Dune, or the warring factions in Mad Fury Road, writers know that when humanity's most precious resource is in short supply, conflict and antagonism follow. But is a post apocalyptic vision from science fiction about to become reality? Well, according to the un, the world has just entered the era of what they call global water bankruptcy. Water systems relied on by 6 billion people and half of the world's food production are being pushed beyond the point of recovery. Now, this marks the first time that scientists have declared water systems bankrupt rather than stressed or in crisis. Put plainly, this means irreversible damage to natural water systems. And that constitutes not just a humanitarian crisis, with the impacts falling disproportionately on those who can least afford it. It's a geopolitical one too. But the good news is experts already know what we can do about it. Welcome back to When Nature Becomes a Security Threat with me, Caroline Lucas. In the first two episodes of this series, we looked at the report the government didn't want us to see, written by the UK's top intelligence chiefs and which warns that climate and nature collapse pose a serious national security threat to this country. And today we turn to the resource at the heart of all of it. Water. Now, water scarcity is often called a threat multiplier. And right now it's contributing to food insecurity. It's deepening economic crises, exacerbating social and geopolitical tensions and potentially fuelling conflict. So in this episode, we'll be asking, why, when does a drought stop being a humanitarian crisis and start being a security threat? Could we really go to war over water? And is the AI race about to make our water crisis even worse? For the uk, the real danger isn't just a tap running dry at home, it's systemic. As we heard last week, the UK leans heavily on climate vulnerable nations to keep us fed system. So when water crises hit overseas, the shock waves reach our supermarket shelves. That's precisely the kind of cascading risk flagged in the second suppressed government report that we've discussed on this series, this time from the Department for Environment, Food and Rural affairs defra on the UK's critical systems. It reveals that water scarcity isn't some distant threat. It's already here. 15% of our surface water and 27% of our groundwater. Aquifers are being drained faster than nature can refill them. And that of course, will only be exacerbated by new water intensive industries like the spread of data centers to feed the booming growth of AI. Now, those data centers depend not only on vast amounts of reliable electricity, but also on large volumes of water, both to keep their servers cool and, and indirectly through electricity generation to power them. Yet global water demand, how much we need, is currently expected to exceed supply how much we have by up to 40% by 2030. So I wanted to ask a question that sounds almost dystopian. Could we reach a point where water is prioritised for cooling servers over supplying people's drinking water? I put that to Ollie Hayes from Global Action Plan uk, who's campaigning to force AI companies to disclose exactly how much water their operations really consume. So Ollie, there seems to be a huge amount of controversy, just about some of the basic facts about how much water these AI data centers really need. So you've got some people saying that the demands have been massively overstated. You've got other people saying they're really underestimated. So why is it so difficult to get accurate data on this?
B
Yeah, it is difficult to get accurate data and partly that's because there is very little transparency from the industry itself. There are no requirements for data center developers or operators to disclose how much water they're using. And even the water companies themselves, some of them claim that they don't know how much water they're supplying to data centers. So we have very poor data, but we do have some good quality research that looks at and tries to estimate the amount of water that the data centres are using. And we know that just a mid sized AI data center will use about as much water as roughly 10,000 people would use. So, you know, a mid sized town or a small town, and obviously what we're seeing is the proliferation of data centers, both in size but also a huge growth in their scale and that inevitably puts water use up.
A
And I guess the picture is really affected by the fact of where these data centers actually are. And I was reading one of your reports, I think that around 84% of proposed sites are in areas that are projected to be water stressed by 2040. And so is that the reason that some of these estimates come up with a very low figure for the amount of nationally that data centers are using in terms of water? Because in fact, once you average it out across the whole country, yes, the figure might look very small, but in actual fact, quite a lot of These data centers are concentrated in water stressed areas.
B
100%. Yes. It matters hugely where the data centers are and where they're taking water from. And we are seeing in this country a massive concentration of existing and planned data centers in the southeast of England, which as we know is the driest part of the country and is severely water stressed, particularly at the moment in times of heat waves. So adding this new novel demand to water supply in these places creates a really acute problem. And as you say, I mean, the industry and its lobbyists like to paint a very different picture. They like to say, well, hang on, per query on ChatGPT, it's just a fraction of a teaspoon, or once you average it out over the course of the entire country's use, you know, it's really not that much. But that's a slightly disingenuous way of looking at it.
A
Yeah, I do find it shocking that as you say, that we don't even know, or even some of these water companies themselves don't know how much the data centers are using. Why isn't there legislation, I understand in the eu, for example, they are bringing in requirements that data centers should report, for example, their freshwater consumption.
B
I mean, there's very poor rules and regulations around data center growth generally. We, a global action plan with our partners at Foxglove, recently challenged the government's decision to approve a data center because there was no environmental impact assessment in that planning application, it was deemed that one wouldn't be necessary. And this is despite the fact that they are demanding enormous amounts of power. Where's that power going to come from? Inevitably, some of it is going to come from fossil fuel generation, which is going to push up emissions and they're demanding a lot of water to cool them. So it's pretty shocking really how little there is in terms of conditions placed on data centers. And we're pushing very hard for those conditions to start to come into effect and for them to be imposed nationally. Because that's the other thing which is happening is that these, these data centers are being looked at on a case by case basis. But actually it's the cumulative effect which starts to get really worrying. And as we are seeing, you know, they are being planned and started to be built in very, very large numbers and the facilities themselves are orders of magnitude bigger than anything we've got at the moment. So the impact is inevitably increasing, but there's very little consideration of it.
A
And one of the factors in all of this that I hadn't fully appreciated was that when the Government designated AI data centres as national critical infrastructure back in 2024. That essentially means, doesn't it, that water companies are now legally unable to disconnect to, for example, AI data centres, even at times of real water stress. And so, in a sense, those data centres are going to be taking water potentially before ordinary householders, for example.
B
Absolutely. I found this genuinely shocking when I realized this was going on. You're absolutely right. The designation of data centers as critical national infrastructure gives them lots of protections in the way that a nuclear power station or certain security infrastructure will be, will be given these protections. But one of the things included in those protections is essentially being insulated from any water company demands to stop using water. So in times of drought or in times of water scarcity, the water company can, in theory, at the moment, say to an industrial user, right, we need to scale back your supply or we need to turn it off altogether. But if data centers are designated as critical national infrastructure, they can't, they can't turn it off. And we know we've done lots of polling of people's attitudes towards data centers and their sense of where they should sit and the priority of many other industries when it comes to scarce resources. And surprise, surprise, they come at the bottom of the pile. People think farming is more important, they think the NHS is more important, they think household use is more important. No one thinks that data centers are worthy of protection in this way. And I think that's going to become a bigger and bigger political problem because, of course, data centers need most water and most cooling when it's hottest and when the rest of us need most water.
A
Yeah, exactly. And I was struck about the irony of the government just very recently launching a new, I think it's a 75 million pound publicity campaign to try to persuade the public to use less water. And so you've got this kind of situation where we are all being urged to take shorter showers, essentially so that AI data centers can go on guzzling as much water as they like.
B
I found this mind blowing, I have to say, and it comes hot on the heels of, I think, about 12 months ago, the government asking us all to consider our email use as a way to conserve. Conserve water as well.
A
Irony is alive.
B
Exactly. It would be funny. It is quite funny. But it would be funnier if it wasn't, you know, also quite serious. You know, we have already an acute problem with water in this country, which is that we are looking at a 5 billion liter shortfall per day in a couple of decades. And those figures are arrived at before even considering data center use. So this is just projections based on the growing population, our existing industries, growing demand, etc. And the changing climate. So this massive additional demand really, really needs factoring in to begin with. And then there needs to be constraints put on the industry. And that's just the opposite of what we're seeing from the government. They think they're going to contribute to a lot of growth, and in that sense they are saying, let's get out of the way, let's let them do whatever they want, but that is just not responsible.
A
And what about technological advances? To what extent is it the case that if we were being a little bit stricter about the kinds of water that they're using, for example, using wastewater or using closed loop systems where they're using less water, how much difference can those technological changes actually make?
B
They can make a difference. And we absolutely should get to a point quickly, I think, where the government must insist that they use the least water intensive cooling processes possible. Now, data centers often don't want to do that because if, for instance, they're using air conditioning or the equivalent of air conditioning to cool their servers, well, that means that there is a significant proportion of their power supply which they can't use to power the servers. And so, you know, their clients get less bang for their buck. And also power is more expensive than water. So there are big economic incentives for the data centers to continue using evaporative water cooling rather than less intensive methods. Whatever they say. But, and this is, I think, the most important point, even if they get to a stage where they genuinely aren't using any water for cooling on site, at least 3/4 of their water footprint, and some studies say it's more than that. Some pushing 90% comes from the generation of the electricity they need to power their operations. So using a lot of water, heating it up, using fossil fuels to turn it to steam in a gas power electricity generating station, that is where the bulk of it comes. And that is why what is absolutely insistent, what should be insistent, is that data centers are only powered by renewables. One, because that is what is required on the power front. Absolutely. We can't have data centers using gas. And we can talk about that a bit more if you like, because they are and they want to. But two, that is the key to reducing the water footprint dramatically.
A
And just the last sort of reflection, in a way, do you think the environmental breakdown itself could become a kind of overlooked constraint on the technological growth here? So for people who are sort of very gung ho about wanting to see the expansion of AI and the data centers and they want the UK to be at the forefront of this race. Do you think it would be fair to say that they are just simply not factoring in themselves the environmental constraints that we have? And therefore, from their own perspective, they should be thinking much more about water use, energy use and so forth, and taking on board some of the kinds of measures that you're talking about?
B
Oh, I absolutely do. I absolutely do. The, you know, the industry began its kind of PR blitz a couple of years ago by saying that this was completely consistent with environmental goals. You know, that this is. We could accommodate this growth within attempts to reduce fossil fuel use and cut pollution. They've changed their tune quite quickly. And, you know, first they were saying it doesn't really use much energy. Then they were saying, oh, well, actually we can just power it all with renewables, which we can't, by the way, there's too much. And now they're saying right out on the open, well, it's climate goals or it's AI. You've got to choose. So they're being unbelievably, sort of brazen about the fact that they don't care about climate goals. The big tech companies who are behind these data centers, ultimate operators of most of the big ones, they've all abandoned their climate commitments because they can't stick to them because of the emissions associated with their data centers. So it was probably greenwash to begin with, but it's now out in the open that they are pursuing a very different goal. And I think the final point on this is that I think the government has a problem here because they have a stated intention to clean up the power sector, get pollution out of power generation by 2030, which is incredibly ambitious anyway, but necessary. And they also have an intention to massively expand the AI industry. And those two things are on a collision course. They just are.
A
There is so much more that I'm going to be covering in this series. We're going to be looking at food security and about the myths really behind the headlines to do with climate migration. So don't miss it. Please sign up@therealDisPolitics.com.
Podcast Summary: The Rest Is Politics – "Data Centres vs. Drinking Water: What Matters To Us Most?"
Episode Overview
In this compelling episode, Alastair Campbell and Rory Stewart (with investigative journalist Caroline Lucas leading the conversation) examine the acute and interconnected crises of water scarcity and the explosive growth of AI-powered data centers in the UK and worldwide. The hosts and their expert guest, Ollie Hayes from Global Action Plan UK, investigate whether advances in technology, particularly data centers, are escalating risks to our most fundamental resource—water—and whether existing policy is keeping pace. They also explore the potential for water to shift from being a humanitarian issue to a full-blown national security threat.
This episode raises urgent, nuanced questions at the intersection of technology, policy, and resource management. Through expert testimony and sharp critique, it's made clear that without a rapid rethink—especially around transparency and regulatory priorities—AI-driven technological ambition threatens to aggravate already critical water scarcity. As Hayes concludes, the UK faces a stark policy dilemma: the ambitions for AI leadership and environmental protection are, under current circumstances, fundamentally incompatible.