Artificial Intelligence relies heavily on data infrastructure especially data centers. At a recently organized series of talks, researcher Sebastián Lehuedé and his colleagues collected insights from grassroots groups and activists around the globe who are opposed to data centers operated by Big Tech companies like Google, Meta and Microsoft. They’re worried about the harmful environmental impact of data centers, which they fear could result in energy instability and water shortages.
Why are activists fighting data centers and what are their main concerns?
The different grassroots groups are active locally and not part of a broader movement. What they have in common are their environmental concerns – specifically the data centers’ energy and water consumption. The management and processing of data requires vast amounts of energy. The data centers built in Ireland, one of the world’s main poles for this type of infrastructure, are expected to consume 27 percent of the entire country’s electricity by 2029. That’s quite a lot. Data centers increase the load on the electricity grid, consuming up to 2 percent of global electricity demand. With data centers becoming increasingly efficient, this might change. At the same time, though, more and more are being built. And then there is the issue of water consumption. Vast amounts of water are required to cool them, which is why data centers in Europe are usually built in colder areas like Nordic countries, where due to lower outside temperatures cooling down the server is easier. In 2019, however, Google planned on building a data center in central Chile, which is home to an increasingly arid Mediterranean climate. According to the first structural design, the Google data center’s cooling system required 169 liters per second, and that in an area where people have been struggling with droughts for years. The average data center uses as much water as three average-sized hospitals. There are also other issues, including air pollution. In the Netherlands, activists worried that the construction of a Microsoft data center would affect their agriculture and, beyond that, ruin the landscape.
Lower emissions, more drought: Is Google‘s data center sustainable?
Google has built a data center in Quilicura, just outside the Chilean capital city of Santiago, which it touts as being one of the most efficient and environmentally friendly in Latin America. In fact, a comparison of Google data centers worldwide based on the company’s self-reported data shows that the Santiago facility produces comparatively low emissions. Around 69 percent of its energy supply is carbon-free, and it emits “only” 190 grams of CO2 equivalent per kilowatt hour. Google encourages customers who are interested in hosting an application via the Google Cloud to choose data centers with low CO2 emissions. At first glance, the data center in Santiago would seem to be the perfect choice. But it’s not that simple. The information provided by Google is self-reported and hasn’t been independently verified by anyone outside the company. Like any other data center, the facility in Chile requires minerals that are extracted in environmentally damaging ways as well as hardware that will eventually need to be disposed of as e-waste. In addition, water consumption at the data center in Quilicura is particularly problematic, given that the region suffers from persistent droughts. Looking at emissions alone is hardly sufficient for assessing the sustainability of data centers. This example in Chile underscores once again how Big Tech companies greenwash their business operations by providing selective information, pretending to offer sustainable choices, when they are in fact aggravating existing environmental crises.
Are activists concerned that data centers located near their homes will have an immediate impact on their everyday lives?
Patrick Brodie from the University College Dublin thinks that new partnerships between Big Tech and renewable energy companies could lead to the exclusive use of renewable energy for data centers, meaning that ordinary citizens are then denied access to clean energy because of green data centers. In Santiago de Chile, people were uncertain about whether there would still be enough water for them with the Google data center being operated there. There were no reliable studies conducted beforehand to estimate the data center’s impact on the local environment and the local communities. In such a constellation, the shift toward sustainable energy becomes a social problem.
Who is responsible for ensuring better protection for local populations – the Big Tech companies, local authorities or national regulatory authorities?
Primarily the Big Tech companies, of course, because they build data centers to profit from them. But we need the authorities to ensure that the Big Tech companies comply with regulatory standards. Alphabet, Meta and Microsoft all issue reports on their electricity and water use, but most of the time, those
reports aren’t fact-checked. That’s why we need third-party actors who keep an eye on how Big Tech enterprises affect the environment and local communities. If this could be done on a global level, then perfect. The problem is that international organizations like UN agencies seem to be quite biased when
it comes to the deployment of technologies like Artificial Intelligence. The UN sees AI as a potential solution to the climate crisis without taking into account that the use of AI itself produces carbon emissions that aggravate this very crisis.
Are the activists you talk to optimistic about what they can achieve? Where do they see limits to negotiating with Big Tech companies at the local level?
In Chile, after years of sustained activism, the people of Cerrillos managed to negotiate with Google, after which the company decided to use a less water-intensive technology. That was a big achievement. But at the same time, lithium is being extracted in the north of Chile, which causes considerable harm.
We need to take a general look at the entire AI lifecycle process to identify problems. But it is difficult to make an assessment because these companies are so opaque. They can’t even say where the minerals that they use to build their technologies come from – probably because they don’t know. Under which labor conditions do people assemble the device? How does the training of the algorithms work? Given that activists and scientists haven’t yet figured out the extent to which the production of digital technologies damages the environment, I would say that activists are rather pessimistic at the moment.
The EU is seeking to establish the AI Act to regulate high-risk AI systems, but the legislation isn’t really addressing the environmental harm caused by data centers and their energy and carbon costs. The EU emphasizes that it wants to formulate AI rules based on European values, so why is it neglecting the environmental impact of AI systems?
This is in line with Western Europe’s historical hypocrisy. Europe is infamous for defending values such as democracy at home while not showing too much concern about human rights in the rest of the world. There is this idea of green technologies, the question is: Who defines what is green? If you want to build
electric cars, the flagship product of green technologies, you need to extract lithium. But indigenous communities in Chile are standing up against the extraction of lithium because the environmental and social harms are so staggering. If the environmental problem isn’t present on European ground, then it’s not surprising that the issue isn’t addressed in European legislation. To a large extent, the European lifestyle depends on the global exploitation of minerals and other resources, and digital technologies are no exception.
Have the activists you know identified any way to reclaim the resources from Big Tech companies and use them for their own good?
Not really. The Chilean activists in Cerrillos didn’t want to address Big Tech companies like that. This is partly a strategic decision because they thought that demanding water justice would be the most effective approach. By seeing Google just as an actor that demands lots of water, they didn’t have to get involved with privacy and other issues. They weren’t even aware of how problematic the situation is on a global level, which is one of the reasons we organized the Data Territories conference at the Centre of Governance and Human Rights in Cambridge. In the case of Santiago de Chile, the local population initially tended to be sympathetic with Google. Due to the successful PR work companies such as Google keep doing, people tend to think: “Oh, that’s great, we’ll get more innovation, and more jobs,” when the construction of a data center in their home area is announced. But through their experiences, I do think that digital rights activists are becoming more and more aware of the nature of these actors. And I’m optimistic that through the actions and the educational work of environmental activists, people will realize just how harmful the influence of their new Big Tech neighbors is. And then we can talk about alternative forms of organizing and alternative technologies. That hasn’t really happened yet, but it might in the future.
Dr. Sebastián Lehuedé is a postdoctoral scholar at the Centre of Governance and Human Rights at Cambridge University. His research focuses on the regulation of digital technologies, with an approach influenced by Latin American critical thought that seeks to decolonialize bodies of knowledge. Lehuedé’s
current project explores the nexus of digital and environmental rights in Latin America. His research has been published in several peer-reviewed journals, including Information, Communication & Society. He also writes for Open Democracy and Progressive International.
The interview was published in the second issue of the SustAIn magazine. For more articles and facts about AI and sustainability, download the whole magazine as PDF here.
Illustration: Kevin Lucbert for SustAIn Magazine #2, CC BY 3.0 DE