The Current - Can deleting emails help save water during a drought?

Episode Date: August 14, 2025

With drought gripping the U.K., the government is urging people to delete old emails and photos to cut the water used by data centres. Civil engineering professor Venki Uddameri explains the link between cloud storage and water consumption, why individual actions have limits, and how to better manage resources as AI and other data-heavy technologies expand.

We are gathered here today to celebrate life's big milestones. Do you promise to stand together through home purchases, auto-upgrades, and surprise dents and dings? We do. To embrace life's big moments for any adorable co-drivers down the road. We do. Then with the caring support of Desjardin insurance, I pronounce you covered for home, auto, and flexible life insurance. For life's big milestones, get insurance that's really big on care at Dejardin.com slash care. This is a CBC podcast.

Hello, I'm Matt Galloway, and this is the current podcast. Caught in the midst of a month's long drought, the United Kingdom has offered its citizens a variety of methods to preserve water. Some of these, you've probably heard before, have shorter showers, don't water your lawn. But the last one is a little more strange. Delete old emails and photos from the cloud. Your Gmail account or Apple Photos app might be basically free, but experts warn the data centers that power these services have unexpected costs. And a big one is water.

It's not just the United Kingdom that's worried. Last week, Tucson, Arizona City Council rejected a local data center project, at least partly due to its impact on the water supply. Tucson, we can all feel how vital and sacred water is. It is an incredibly precious resource. and, you know, Beale wanted to use millions of gallons a day. That's completely unacceptable. Thank you, Damari, is a professor of civil engineering at Lamar University who studies data centers and water management.

He joins me from Beaumont, Texas. Good morning. Good morning. So you heard there the UK recommending people delete old emails and old photos. What do you make of that suggestion as a way to conserve water? Right. So it is a short-term fix. for their imminent drought that's happening there. The success, of course, depends on a large number of people actually taking their advice and doing that. Otherwise, you know, the per person impact is very low.

So if not many people do it, then it won't be very effective. Right. How much water does an average cloud storage data center consume? Give us sort of walk us through how that works. Yeah. So a small data center would serve about 2,000 people, but typical data centers that we have, especially the bigger ones, the hyperscale data centers. You can think about it as the amount of water that's required per day by a community of, you know, 30,000 to 50,000 people. So it's substantial. But cloud storage, is that really the main culprit? I mean, shouldn't they be asking people to

give up chat GPT for a while? Yeah, all of that is, you know, certainly there. So they're trying to say that, you know, you're having this data that's just sitting there. You're probably not seeing it for a while. So they're probably saying that there is going to be a less impact to the persons rather than give up chat GPT for a while. Although, of course, you know, the amount, you know, the chat GPT, giving up chat jeopardy for a while is certainly gets you more water. gains than cloud storage, but it's not something people might give away. Yeah, it's interesting because, yeah, it seems to me as though if they are actually

serious in tackling the data center problem, there are other ways to do it. Chiefly, maybe why not sort of talk to the companies that run these things? Is there not some message they could be sending to the governments who allow it and to the companies who build these things? Absolutely. So those are all long-term fixes that are, you know, the biggest amount of water uses comes from cooling these systems and then, you know, keeping them running. And there are other ways to cool it other than evaporative cooling using water. People are looking into it, but the technologies are expensive, which means eventually that's, you know, it's going to take some time. So this directive from UK government is simply, a short-term fix, it certainly doesn't do much in the long run because that very idea of a cloud is you want to store your data reliably as opposed to putting it on a jump drive, you know, that might go back. Give me a sense of how much more power it takes to run chat GPT, for instance, versus storing something in a cloud because there is a difference there. Yeah, absolutely. There's a substantial difference about 5x. You know, if it takes one unit of energy to store, this will probably take about, you know, five to six times at least. I mean, I'm not talking about like imaging and some of the modern things that you can do, but certainly just text-based searches, just like Google-type searches, for example, or X-based searches.

And what about emails? How much does an email? that is sitting there in the cloud, how much does that consume water potentially? And it's hard to extrapolate, I think, exactly one email, but maybe you can for us. Right, right. So a typical email without any, you know, big attachments takes about 30 milliliters of water per day. So if you have big, you know, you could talk about 100 millimeters, 50 to 100 millimeters, if you have attachments and things like that. But, you know, if you look at somebody's email storages, particularly Gmail or Yahoo,

which gives us quite a bit of storage space for free, you're talking about hundreds of thousands of emails per person. So all of that can add up if they delete it. And it costs energy even if I'm not looking at that photo or document that's stored in the cloud if it's just sitting there. That costs us power? Yes. Yes, absolutely, because the computers still are doing a lot of things internally to make sure that they're swapping memory, they're moving those files, because sometimes to make up more memory in a certain location, there is antivirus software that's running in the back, but make sure that this is all there.

So, yes, there is a lot of internal operations that are happening, even though you're not. looking at it. So what do you think we should do about this conundrum, that we're all racing towards AI and data centers, but we're also dealing with climate change and forest fires and heat waves. What do we do? Yeah. I mean, this is, this hits very close to heart to me, because I use a lot of computations to study water conservation. So, you know, I've been dealing with this conundrum, which is what got me into this research. The key issue is, the, the, the key issue is, is, you know, rather than look at, there are two big aspects, I think. One is the energy and the water nexus that's there.

And for that, you know, we need to move more towards renewables or solar, wind, other types that don't use as much water that have a much smaller water footprint. That's one aspect. The second aspect is do we do evaporative cooling or do we do some. or combination of evaporative cooling with water versus other ways of doing it. So people are looking at configuring networks differently, data centers differently. And then the question also becomes, do we do big, hyperscale data, a few big hyperscale data centers vis-a-vis relocating some of these data centers to places where there is more water or balance it?

But that comes at a computational cost because now you're moving in the electronic space a bit more to capture that data and bring it to your computer. So it's not a perfect solution, but certainly there are solutions. There's no perfect solution per se that costs more money or it costs you more in terms of resources. So that's a very big active area of research. I think this is more, you know, asking people to delete emails is more a short-term fix. It can not be a long-term fix. And do you think people will buy into that on this kind of scale where it actually makes a difference? I don't think, you know, personally I was putting myself in that position and seeing what would I do.

And certainly, you know, there is certainly some email hygiene that we all can use, like deleting spam and, you know, mails that we are never going to look. But if you ask me to delete some emails, you know, that are important to me, it would be a difficult choice, you know. So it's a question of, again, you know, the big issue that boils down to is, is more. my deletion of this email really going to make a dent. And unless there's a collective feeling that it will, this will not happen. You know, I've talked to people about this before and concern that the hyperscalers just have, like Google and what have you, have so much money that they're not as concerned about being energy efficient as maybe some of the smaller ones who have to work within their means

and in the markets where they know people directly, maybe. But like, are these big hypercenters, are they moving to be more efficient? They are certainly, there is definitely a push to be more efficient, especially as the price of water goes up. You know, one of the big problems with water is we don't price our water correctly. And there is a good reason for doing that because it's an essential commodity. But it doesn't mean that, you know, the price of the water for a water center has to be the same as the price of water for human consumption. So there has been push, as long as we have that driver, it will push towards efficiency.

But water historically has been very cheap or almost free, and that's been a big cause. I'm curious, Venki, you have your own conundrum. You said a lot of your research has done, of course, using what you imagine is happening in data centers, of course. So how do you come to terms of it for yourself? Well, you know, it's a tough one because, like, you know, predicting climate change, predicting what happens to water, predicting droughts. It requires a lot of computations. And but at the same time, you know, the idea is, you know, it's like using cars. We got to use cars because, you know, we have no other choice, but we have to use it responsibly, right?

So try to, like, you know, when I run my computers, I don't have a lot of background processes running, and then I have dedicated computers that I try to run at nights because it just is not, well, there are not many other users competing for it. So there are things we can do or have to try to do. But certainly, you know, there's no perfect answer at this point. Okay, it's interesting stuff and a conversation, I'm sure that's going to be happening for quite some time. Vanky, thank you very much for this. Thank you. I appreciate it.

Thanks. Thank you. Tomari is a professor of civil engineering at Lamar University who studies data centers and water management. This has been the current podcast. You can hear our show Monday to Friday on CBC Radio 1 at 8,000. 30 a.m. at all time zones. You can also listen online at cbc.ca.ca slash the current or on the CBC listen app or wherever you get your podcasts. My name is Matt Galloway. Thanks for listening. For more CBC podcasts, go to cbc.ca.ca slash podcasts.