Jacob Tompkins: Digital water footprint

Michelle: Hi, I'm Michelle Brogan and welcome to today’s Allianz Water Expert Podcast.

At Allianz, we’re committed to raising awareness about global water challenges and supporting positive change through education, innovation, and collaboration. That’s why we created the Allianz Water Hub, a platform where businesses and individuals can learn about the value of water, the risks we face, and the practical actions we can all take to protect this essential resource.

Michelle: Today, we will be talking to Jacob Tompkins OBE, Managing Director of The European Water Technology Accelerator. Jacob is Chief Technology Officer at The Water Retail Company one of the new water companies in the UK, co-founder of Thunderstorm, an agency that links sport and entertainment with environment, and a visiting Professor at Exeter University. Jacob was also the co-founder and director of Waterwise, a UK NGO dedicated to promoting water efficiency and conservation.  Jacob is also an Officer of the Order of the British Empire (OBE) for his services to water conservation.

In this episode, we’ll be discussing a fascinating and increasingly important topic, the hidden water use behind our digital lives. From Taylor Swift downloads to ChatGPT searches and even cat videos, data centres require huge amounts of water for cooling. This water use is rapidly growing as data demand explodes and is a major concern, especially in water-scarce regions around the world.  

Jacob will also address key questions such as the impact of data use on water consumption, why some regions are more vulnerable, and what lessons we can learn from initiatives like the proposed ban on water use in data centres in Utah.

Welcome, Jacob.

Jacob: Wow, that’s an introduction. Thank you so much! I need you to do my warm-up all the time.

Michelle: I can't wait to get into this conversation. It's such an interesting topic, particularly for me, because I work in a in a large corporate, Allianz. And our business is run on data centres. So, I can't wait to learn more.

Jacob: I'm not singling out Taylor Swift. She is an excellent individual. It's any kind of downloads on Spotify. A ChatGPT search, ironically, people often search for things like saving water. Ten searches, takes one litre of water, and then if you're streaming full Taylor Swift concert for two hours, that's 1,000 litres of water. I know the other bands are available. That's Tom waits, Pink Floyd, Taylor Swift, anyone? And then the old cat videos. Every minute you spend on Tik Tok, that's a quarter of a litre of water. 

So, if you think about the way on the bus, you see everyone either on Spotify or on TikTok, all of that is using water all the time. These things are amazing in our lives. We use them a lot, but I don't think we think about the consequences that are associated.

Jacob: Okay, so if you think about a data centre, just huge racks of computers effectively in a big thing that you'd see on like a film all storing data. Now, if you think about how hot your laptop gets or your desktop, the amount of heat that's thrown out of that. Multiply that up millions and millions of times the amount of heat that's being thrown out by these machines that are storing all of our data. 

Now, in order to cool that what they generally do is, they use evaporative cooling. So basically water, because it has a very high specific heat capacity, can absorb a lot of heat. And you can tell that when you when you switch the kettle on a home. And if you've got an energy monitor, you can see that dial shoot round. So heating water and cooling water is very expensive in terms of energy. So, what happens is the heat from the data centre goes into the water that then evaporates off and you lose that water. So, in order to cool the data centres, we’re using huge amounts of water.

Michelle: Okay, that makes sense.

Jacob: Okay, this this is interesting because we're seeing more and more data centres. The majority are in the US still about 90% of them. But we're now seeing data centres, in places like Mexico, and we're seeing them in lots and lots of other places that are water stressed, including, ironically, the Thames Valley, for instance. 

So, there's a lot of data centres proposed in the UK. And by 2030, the amount of water used by data centres will be the equivalent of the amount of water used by the city of Liverpool. So, we're talking about a huge amount of water. And then if you're looking at where they're concentrated, particularly in the US, they quite often in desert states. So, in Reno there are tax breaks. So, Reno now almost half the city's data centres, there are huge data centres in the desert. So, it's to do with where we're placing them and the concentration of those data centres. Data centres like to herd. So, they tend to be next to each other for management purposes, for energy supplies, that they'll build a new energy grid to supply those data centres. So, we tend to get clusters of them. And the impact of that on a catchment or a watershed is enormous. It's like suddenly building a massive new city.

Michelle: Okay. How are they getting water out to the desert?

Jacob: So, in Reno it is taking water. Well, that's a really good question. It's taking water from the municipal supply. So, you can immediately see there are going to be conflicts between the water for the data centres and the water for the towns. And also, the water for cooling power stations, the water for growing stuff. It's a major challenge and it’s not just data centres. 

I mean for instance, Elon Musk's mega factory that he was going to build a battery factory they're going to build in Berlin was going to use a huge amount of water. That was stopped by the Berlin authorities. So, we're going to get to the point where the clash between really heavy water intensive industry and things like data centres, they're going to clash with water for agriculture, water for human consumption, and water for the environment. I mean, the impact on the environment is huge.

But yes, the question you've asked, how are you going to get all that water to this in the desert? Strangely, these questions aren't being asked. It's where do we get our tax breaks? Are they going to build us an energy grid? There's a lot of space. There's a lot of space in the desert, and it's nice and flat. And if you want to build a huge shed to store data, there you go. Unfortunately, it's also very hot so even more cooling.

Michelle: I can see the problem there.

Jacob: I'm not blaming the people who do the data centres, and I'm not blaming the companies who do it. It's all of us. You know, we've on our phones, we've all got loads and loads of apps, all of which use data every time you send an email and then you send a group email and you say, yeah, thanks, I've got that. And everyone else says, thanks. And you say, yeah, and you send a smiley face or a thumbs up each time you do that, that uses water. And we don't think about that. And really, we shouldn't have to think about. That's another environmental guilt. 

But the data impact is huge and people don't understand this. They don't understand that music downloads and cat videos and all this all contributes to a localised water crisis in a desert, in Reno somewhere, or in Mexico or in LA, in other parts of Latin America, in the Middle East, everywhere. It's becoming a growing. I don't like to use the word ‘a problem’, challenge because I'm sure we'll solve.

Michelle: Yeah. I think, you know, we hear people talk about it in the public domain. And, I mean, we hear the big facts, but we don't really understand.

Jacob: No, people talk about data centres, data used, blah, blah, blah, but they don't say your cat video, which is very funny, is using this much water. And that means that that data centre near Reading or near Reno is using this much water and that is impacting on the amount of water for the river or for humans to use or for other industries. And it's never laid out like that. We always talk about things like biodiversity instead of just plants and animals. We need to spell this out so people can make the choices.

Michelle: Yes, exactly, exactly. It all comes with education. 

Jacob: Which is why the fact you're doing these videos is great. Quick plug for the Allianz video.

Michelle: That’s exactly the reason, because it's a topic that I don't think a lot of people in the public domain really understand and understand how they can make changes in order to address the issues.

Jacob: Scientists and environmentalists love to talk to each other in jargon. And then the stuff that comes out at the end of that is just scary. There's a global water crisis and it doesn't empower people but explaining it to them properly and then telling them what they can do, that's much more empowering. And we've got that gap between science and the public. And that's why we need more storytelling and more narratives, which quite often used up data ironically.

Michelle: Yes, definitely. We do.

Well, that's what we're talking about. So, there's a fantastic graph by the US Geological Survey and it shows the Earth, the blue planet. So, if you're looking at covering water, but actually if you take that water off the planet, so the planet's like that big and you put that into a tiny sphere, there's only about that much water on Earth. And then if you think about the water that either isn't salty or isn't locked in deep, deep aquifers or frozen. It's a tiny bit. And then if you think about the available bit. It's a minuscule amount of water that is available for power generation, agriculture, industry, human use, and nature. 

And if we were on a rocket ship flying through space, which actually we all are, and there was a room with all the water in and it was finite, we wouldn't be doing these things. We'd say, hang on a minute, there's only three bottles left in the water room. And this is the difficulty that we seem to think it's limitless. And people say it's the same water recycles round. But the difficulty is the way we're messing with the climate and the way we're consuming so much, it is actually a finite resource. 

And the difficulty we're having now is there’s too much water. So, you know, the catastrophic floods we saw in Valencia. Too little water. So, the drought that we've got in the north of England, for goodness sake, at the moment. All the water is polluted. And we're seeing this around the world. We're seeing microplastics in sewerage. We're seeing PFAs. 

So not only are we depleting the water and messing with the environment, so the water moves around in a different way, we're also polluting the little water we have. So, we need to look at water use of all of these new industries. It's not just data, it's everything and every way we use it. When we think about industrial strategy or growth, we need to consider water in that. As I'm saying, if we were on a spaceship and there was only limited amount of water and someone suggested, let's have a Jacuzzi, people might say, well, hmm hang on. 

There are challenges for who gets the water first. And then you come into something called cascading failure. So, there was a really interesting example in France in 2007 I think it was where the water level in the rivers had dropped really low. And there was a choice between giving water to the town or giving water to the nuclear power stations. 

And the difficulty there was if you don't give water to the power stations, you don't have the power to supply the water purification systems so the failures cascade. And when you see low water levels, it also means that the cooling intakes for power stations fail. So, you can get a failure in water that leads to a failure of power that leads to failure in agriculture. All of these things are linked together. 

So, if you suddenly throw in another big thing like data centres, it messes with that balance. And the balance is pretty fine already.

Michelle: Yeah, definitely and with the growth in digital consumption, it's not really helping matters.

Jacob: Well, the proportion of water being used by data centres is very, very large and growing. So, what we will see is conflicts, as we've said, between water for agriculture, water for businesses, water for domestic use and the environment. And the one that will probably give out of all of those is the environment. So, we will trash the environment even more if we're not careful. 

There are all solutions. But we can't just ignore the water and energy impacts of data centres and that's not telling people to stop watching cat videos. It's just saying we need to be aware of how much that uses and address how we are going to do this in a sustainable way.

Jacob: Okay, now that's the interesting bit. So first of all, you can change the cooling method. So, you can use a refrigeration-based method or you could use an air blowing method. I mean some refrigeration uses water as well. But there are methods that you can use that don't use water. The difficulty with those is they use a lot more power. So, you're balancing water against CO2. So that's a difficult trade-off. 

What you can do, and the first thing I would suggest is green coding. And that is when you write software. A lot of software engineers, as there's more and more computing power, they make their software more and more complex because you've got like a supercomputer, well then you can run it as many times as you like. You don't need to worry about the capacity, just run loads and loads. But what you're now having in the green computing movement is they are optimising their algorithms to reduce the resource consumption of that data and those algorithms. So, they're optimising the way data is stored. They're optimising the way algorithms run and that means a lot less energy to do the same thing. So, you can still do your Taylor Swift downloads and cat videos, but they're using ten percent of what they were using before. 

So green coding definitely we need and it may cost more, but also green coding is efficient. It's elegant. It's much more beautifully done and it's about efficiency. Like we talk about efficiency in energy or water, we need efficiency in coding and we need efficiency in data storage. 

And then when you get to the physical data centres there are aspects about where we locate them. So, if we locate them in colder climates there's less cooling. If we locate them in places with more water, then there's more abundance so you're not messing up that catchment. Or some people have been looking at putting them under the sea where the ambient temperature is much, much lower and you're not taking up the land. 

And then the other options are offsetting. You know, offsetting sometimes has a bad name, but both IBM and a couple of others and Amazon have been doing this. So, what they've been doing is they've been working with water utilities and technology companies. And they've been saying we're going to build a data centre in Paris. But what we're going to do is we're going to work with the water utility in Paris to reduce leaks, for instance. So, they would go in and they would say the data centre uses twenty megalitres. We're going to fix twenty megalitres of leaks. So, the impact of the data centre is no new net water. 

Now you have to be careful of double counting and you also have to be careful that it is water neutral within the catchment. So, for CO2 it can be water neutral anywhere in the planet. Water it's very localised. So, you can't say I'm going to fix some leaks in China for a data centre in Mexico. It has to be within that same watershed or that same aquifer underground. Rock water sort basically. So that's an interesting approach. 

And then the other one is recycling and reusing the water. So, when I was saying before, a lot of cooling is done through water evaporative cooling. So, you've got a hot data centre, you put water on it, it all evaporates, it cools it down. Oh, hang on, we could catch that evaporating water and re condense it and then reuse that in a cooling cycle. And there's quite a lot of this. And I know Amazon I think is pledged to have 100% water reuse by 2030. Again, not that much clarity on whether that is globally or whether it's at a localised level. 

Some of the big data companies have also talked about buying recycled water. So, they would go to another industry, and this is called industrial symbiosis. So, let's say you've got a big industry next door that's washing vehicles. And they have a lot of low grade, slightly polluted water. Well, you could potentially use that. So are there are different options there. 

And then the other one is we could all use less data but that's unlikely to happen. And the other one is reuse the heat. And that again depends on location. So, you could be using the heat for district heating systems they're complex to do particularly well done in Scandinavia for instance. But I haven't seen them use that much yet with data centres. Okay. But you know, they have a nice swimming pool located next to every data centre and use that heat to warm the water.

Michelle: Do you know a way of catching the evaporated water?

Jacob: Well, that's the idea. So, you can recondense it. So, you've got the evaporated water and then you can condense that water. So, it then runs down a bit like sort of holding a plate above a kettle. So, you condense that water. Obviously, it's still warm and there's an issue with the way the condensate work. But yes, some data centres are looking at that and trying to recycle in the loop.

Michelle: Yeah, okay. Well, that sounds good.

Jacob: It's both. It's always both. It's always very easy to say to blame big business and we all love blaming the big businesses. But actually, we're the people that consume the stuff. So, you know, there's no point on going on to ChatGPT and saying, how evil are these data companies.You're the one using ChatGPT. We're all using that data. 

So, several things you can do. You can think about the amount you do. You can think about do you really need to send that email a bit like when people say, do we need to print this email off? Maybe you really need to send the email. So, think about your data use. But that's very, very difficult. 

The other bit is to use green software. So, to specifically search out green technologies, green software, green data storage, there are options for that. You'll pay more as always, but it will help transform the market. And there are potential for data offsets. 

And also ask your data centre: What do you do? Not your data centre. Your service provider. What are you doing? How is your data stored? All those sorts of questions. 

So, this is a shared issue. Data centres are not standalone. They're not doing this just for the fun of storing data. They’re storing our data. So, it's all of our responsibility. And you know, if we want cheap anything then someone has to pay for it. So, cheap data is being paid for by very high water and energy consumption.

Jacob: Okay, so some of the big tech companies like Microsoft and Amazon and others are looking at working with water utilities, municipalities and tech companies to offset their water use. So, the data centre are uses 20 megalitres of water. 

What they can do is work within the catchment or the watershed and find other businesses that are using a lot of water to help them reduce their leaks. Or it could be, you know, you're from Allianz, so they could be working with, an insurance company helping houses reduce their leaks. So, what that is doing is reducing insurance risk, reducing damage to houses, reducing the leakage and offsetting the amount of water the data centres are using. 

So, what the data centre would say is, okay, we're using 2,200 megalitres or whatever. We will then offset that by fixing leaks within the same catchment. Now you have to be careful you're not double counting. You have to be careful within the same catchment. Lots of things you have to look at. 

But there are some initiatives like this and that's really quite good because it's within the catchment of the data centre and they are things like fixing household leaks also contributes to reducing household bills. 

But that's good stuff. We need to see more of that done and we need to see more quickly. And we need to sort of see that standardised.

Michelle: Yeah. Okay. Very good.

Jacob: They haven't managed to do it yet. The legislation is still going through. It's still being debated. It's been interesting because the less water you use, the more energy you use. So, there are alternatives to water cooling of data centres, but they use a high amount of energy. So, they have to be careful that they're not offsetting one environmental problem with another. There are also issues that if they do this, they may push the data centres to other states where it's cheaper and they lose revenue. 

But the positive issues are what it has done has focussed a lot of attention on how you can make water efficient data centres, how you can cool data centres in a different way and it's focused the mind on that problem. So, in my view, this can be solved through consumer action. As I said before, green software legislation and the push for legislation and the engagement of legislatures around this, and also the tech companies themselves and new technologies coming in. 

So, there isn't a single ‘let's make it illegal’ or there isn't a single ‘it's all Amazon's fault’, let them sort it out’. It has to be everyone acting together in concert. Like most problems really.

Jacob: Okay, well this comes back to what we touched on before. So, you can use energy based. But again, if you're not using water that's generally higher energy so airflow all that sort of stuff. As you said you can put data centres in cooler places, which therefore are cooler. That's fairly obvious but then all the data requirements in those places, what about moving the data around? What about access, physical access to those places? There are other issues there. 

And then there are some people looking at putting data centres under the sea. Again, some issues around access, but quite interesting. So yeah, there are lots of people looking at data centres and there's also an issue of scale of data centre. Is it better to have a big data centre, small data centre, distributed storage, concentrated storage. And then there's a multivariate approach to that because costs are different. And again, it links to things like tax breaks, land availability. 

So, it's complex. But there is a lot of work being done on this, except the public don't seem to realise, people don't understand. But there's a lot of work going on things like cooling data centres.

Michelle: Yeah, I think so. But thank you so much for joining the podcast today. It's been such a lovely, conversation and I've learned so much now, and I have questions to go back and ask my organisation to understand what we're doing about water. But thank you so much, for joining.