Category
🗞
NewsTranscript
00:00 Let's say I ask 10 to 50 questions of ChatGPT.
00:04 The hugely popular chatbot would need to gulp about 500ml of water to provide me with the
00:09 answers.
00:10 That's roughly the amount in this standard bottle of water.
00:14 Hundreds of millions of users then, all submitting prompts, quickly shows just how thirsty artificial
00:19 intelligence can be.
00:20 It's little known details like this that help to shine a light on Big Tech's water
00:25 footprint - an often overlooked issue, but one that is increasingly coming under public scrutiny
00:30 as global demand for AI gathers pace.
00:33 In July of 2023, for example, protesters took to the streets of Uruguay's capital city
00:38 to demonstrate against Google's plan to build a data centre.
00:42 The proposal sought to use vast quantities of water at a time when the South American
00:46 country was suffering its worst drought in 74 years.
00:50 Xiaolei Ren, a researcher at the University of California, recently published a paper
00:55 looking at the resources needed to power AI models, as well as what can be done to try
01:00 to quench the thirst of this juggernaut.
01:02 In general, the public is getting more knowledgeable and aware of the water issue.
01:06 If they learn that the Big Techs are taking away their water resources and they're not
01:12 getting enough water, nobody will like it.
01:14 I think we're going to see more clashes over the water usage in the coming years as well.
01:18 So this type of risk will have to be taken care of by the companies.
01:23 In just the last few months, tech giants deeply involved with AI research have reported a
01:32 massive jump in their water consumption.
01:35 Researchers say that AI is partly to blame, as companies use millions of litres of water
01:39 to keep temperatures in check at their data centres.
01:43 It's easy to think of the internet as something that's intangible, but whether it's asking
01:47 questions of ChatchEBT or logging in to check your emails, the computing power needed to
01:51 keep these websites running relies on a lot of water.
01:55 Microsoft disclosed in its latest environmental sustainability report that its global water
02:00 use rose by more than a third from 2021 to 2022, climbing to nearly 1.7 billion gallons.
02:07 For Google, total water consumption at its data centres and offices came in at 5.6 billion
02:13 gallons in 2022, a 21% increase on the year before.
02:18 It means that Microsoft's annual water use was the equivalent to more than 2,500 Olympic-sized
02:23 swimming pools, while for Google, this would be enough water to fill nearly 8,500 pools.
02:29 Both companies are working to reduce their water footprint and become water positive
02:34 by the end of the decade, meaning they aim to replenish more water than they use.
02:39 It's notable, however, that their latest water consumption figures were disclosed before
02:44 the launch of their own respective ChatchEBT competitors.
02:47 The computing power needed to run Microsoft's Bing Chat and Google Bard could mean more
02:52 water usage over the coming months.
02:55 Water scarcity is a big problem on every continent, and one that's being exacerbated by the climate
03:00 crisis.
03:01 Here in Dubai, a region where the heat is extreme and water is precious, policy makers
03:06 and business leaders are gathering at COP28 to discuss what more can be done.
03:10 What really needs to be in place is an international governance body that can regulate, and that
03:16 is what COP is to a certain extent, a global governance institution that tries to regulate,
03:22 that tries to monitor complex problems.
03:24 One subset of that complex problem is AI, which is at the moment in a very transient
03:31 phase.
03:32 It's a reflection moment for AI, so it could go either way, and it's very important that
03:36 we govern that direction to be less energy intensive and more within planetary boundaries.
03:44 Data centers are part of the lifeblood of big tech, and a lot of water is required to
03:49 keep them running smoothly.
03:50 For Meta, it's these data centers that generate not only the highest percentage of its water
03:55 use but also the lion's share of its energy use and greenhouse gas emissions.
04:00 Tens upon rows of computing servers in hubs the size of warehouses work to store and transfer
04:05 all our digital information, with almost all this energy then converted into heat.
04:11 Companies like Google, Amazon, Meta and Microsoft all use various cooling systems to prevent
04:16 their servers from overheating.
04:18 One common method involves the use of cooling towers.
04:21 This heat rejection device works like a giant air conditioner, typically using water in
04:25 an evaporation process to expel building heat into the atmosphere.
04:30 Another technique often used is outside air cooling, or free cooling.
04:34 This system takes the outside air, which has a lower temperature than inside the building,
04:38 to directly cool down the servers.
04:40 This process can work without the need for cooling towers, but water evaporation is still
04:45 needed when the outside air is too hot, often at the height of summer when people use water
04:49 the most and drought is at its worst.
04:52 Basing data centers in cooler climates is one way to get this free cooling, but at present
04:57 only a fraction of them are currently housed in colder environments.
05:00 The US, by a long way, operates far more data centers than any other country, with more
05:05 than 5,300.
05:07 That's 10 times more than second place Germany.
05:10 Anne Skeet, Senior Director of Leadership Ethics at the Markkula Center for Applied
05:14 Ethics says tech companies will need to be a little more judicious on AI if they are
05:19 to meet their environmental goals.
05:20 I think we are going to see some activity from regulators.
05:23 There's already some increased requirements coming in the new year for companies to disclose
05:29 their water usage in the European Union.
05:31 And I think that requirement is a necessary one for people to start to appreciate the
05:37 issue and think about their own choices in it.
05:42 They may have to become a little more heavy-handed in terms of what they're asking companies
05:47 to do and even, you know, think about placing certain limitations on companies if we don't
05:53 see either market forces or innovation contribute to bringing the use of water down.
05:59 AI is one of the most prominent and fastest growing workloads in data centers, which means
06:04 big tech is likely to need a lot more of this precious resource.
06:08 In fact, the computing power required for AI is estimated to double every 100 days and
06:13 it's expected to increase by more than a million times over the next five years.
06:18 Wren and his colleagues have predicted that global AI demand could be accountable for
06:22 up to 6.6 billion cubic meters of water withdrawal by 2027.
06:27 Remarkably, that would be roughly equivalent to the total annual water withdrawal of Washington
06:32 State.
06:33 Researchers say this is deeply concerning at a time of growing water scarcity.
06:37 Alongside the urgent environmental risks, tension over the water usage of data centers
06:42 and human needs has been at the heart of social problems across the globe.
06:47 Following the protests in Uruguay over Google's plan to build a data center, the tech giant
06:52 reportedly said the project was still at an exploratory phase and stressed that sustainability
06:57 remained at the heart of its mission.
06:59 It's thought social challenges such as this are likely to increase over the coming years,
07:04 prompting many to ask what some of the solutions could be to meaningfully address big tech's
07:08 growing water footprint.
07:10 This water is a shared challenge by all of us, so nobody can solve this issue without
07:16 global efforts.
07:17 So as users, if let's say we're just doing some random stuff, we might want to avoid
07:24 using generative AI because those are generally more resource consuming compared to a simple
07:31 search.
07:32 So there are a lot of ways individual users can contribute to the water users' saving.
07:37 But the first step is that we need to be aware of this issue.
07:40 We need to have the visibility of the water efficiency of the services.
07:44 [MUSIC]