AMY GOODMAN: We are broadcasting from Sheffield Live! community television and radio in Sheffield, England, from the same building as the Sheffield DocFest, where Steal This Story, Please!, the documentary about Democracy Now!, is premiering here. It moves on to Docs Ireland. We’ll be in Belfast next week.

We begin today’s show with how the AI race is impacting the environment as Big Tech companies like Amazon, Google, Meta and Microsoft, along with other AI players and governments, are racing to build the massive data centers that power artificial intelligence. We hear a lot about what AI technology might mean for society, and much less about what this global infrastructure looks like as a material system with measurable environmental impacts.

A new investigation by U.N. scientists aims to change that. The report warns AI’s water use in 2030 will match the needs of 1.3 billion people, while its power use will be triple that of Pakistan, Bangladesh and Nigeria combined — countries with a combined population of 650 million. Already in 2025, global data centers consumed an estimated 488 terawatt-hours of electricity. If treated as a nation, they would have been the world’s 11th largest electricity consumer, behind France and ahead of Saudi Arabia.

The report also frames this as an environmental justice issue. Ninety percent of AI computing is concentrated in the U.S. and China, while the rest of the world, as well as communities within those countries, disproportionately bear the costs — minerals extraction, e-waste, water shortages and more.

The investigation was conducted by the United Nations University Institute for Water, Environment and Health. It’s titled “Environmental Cost of AI’s Energy Use: Carbon, Water, and Land Footprints.”

We’re joined now by Kaveh Madani. He’s the director of the institute and a co-author on the report. He previously served as the deputy head of Iran’s Department of Environment and earlier this year was awarded the Stockholm Water Prize, often referred to as the Nobel of Water.

Welcome back to Democracy Now!, Kaveh Madani. If you can start off by summarizing what you have found?

KAVEH MADANI: Thanks for having me, and thanks for your interest in this very important topic.

So, what we know is that we are facing the fourth industrial revolution, that is changing our lifestyle, the way we work and we live. And when it comes to AI, most people understand AI as a digital technology, as a virtual thing, as something that is in the clouds and up somewhere out there. What we tried to do in this report was to remind people that there’s some physics to all of this, and there’s a massive infrastructure and material supply chains that back this service or this innovation. And if we want to understand our impact, we have to think about that very long supply chain, that starts with the extraction of critical minerals in the poor areas of the world and leads to the manufacturing of all the technologies and the hardware, then construction of the data centers and other infrastructure, then operations of that infrastructure that requires massive amounts of energy and has major footprints, and then, at the end of the life cycle, we have to deal with the e-waste.

In this particular report, we focus very much on the operations and the massive energy use of the operations, meaning that our day-to-day interactions with AI tools result in a lot of energy use. And producing energy, as we know, not only comes with carbon footprint, but, as we remind in this report, also with water and land footprint and other ecological impacts that often we don’t pay attention to, especially when we keep talking only about the carbon element. We wanted to remind people about those local impacts, what is happening.

And then, when you look at the numbers, you see massive, massive numbers, major numbers, enormous use of energy, and that means that we have to deal with a major challenge. If we continue using AI in the way we are — and it seems that that would be the case — then we would need to satisfy the power hunger of AI. A lot of investments are happening in renewables, but, as we discuss here, if, you know, the increase in power consumption results in the stopping the retirement of fossil fuels, the decarbonization process would be halted and compromised, and that’s not something good for the world.

Yet, we clarify in the report that AI is not bad. AI is a technology on its own. It’s like a knife. You can save a patient’s life in the operating room with it as a doctor, or you can kill people with it as a murderer. The way we use AI would determine if this is going to be a good technology for humanity or not. And we say that we have to proactively manage things and think about those impacts, if we want this revolution to be sustainable and fair.

AMY GOODMAN: I wanted to read you a statement by your colleague, the lead author of the report, Dr. Miriam Aczel, who said, quote, “What surprised us most is how often the choices that look greenest from a carbon perspective end up worse for water or for land. If we keep judging AI sustainability by carbon alone, we might think that renewables make AI infrastructure clean but that is solving one problem while creating other problems, often in places that didn’t ask for it.” Can you expand on this?

KAVEH MADANI: So, yes, the climate change discussions and discourse around the world is very much focused on carbon, and that’s something very normal. But we have shown in our research in the past that some of the renewables, although they have — they produce less carbon, when it comes to their water impact and land impact, they have massive — they have big water footprint. For example, biofuels, on average, are 70 to 400 times bigger in terms — you know, have a water footprint that is 70 to 400 times bigger than some traditional fossil fuels. So, if we go to a greener choice or a, you know, carbon-less choice, it doesn’t always mean that there are not other impacts. And we have to understand these tradeoffs when we deal with them.

Hydropower is another one that we sometimes brand as a clean energy, but we know that building a hydropower dam results in a lot of ecological footprint. Reservoirs also have high evaporation rates and lose a lot of water, so that means that they have high water footprints. So, what we try to remind here is that, first of all, you have to choose the energy sources that are suitable for each community. Hydropower might work in Brazil or Canada or Norway, but it’s not necessarily the best choice for California or Iran, where I come from. So, that’s something to keep in mind.

But also, we have to pay attention to the fact that now we are adding more energy demand. Post-Paris, one of our challenges — post-Paris Agreement, one of our challenges has been to decarbonize and reduce power consumption. So, we keep adding renewables, but at the same time we are increasing consumption. That means that we are not making progress. Decarbonization would be only successful if, at the same time that we are increasing demand, we also — you know, we also retire the fossil fuels that we rely on. That means that we have to produce more and more renewables, invest more and more in renewables, but also not every type of renewable, the renewables that are cleaner, and then think about where we are placing those, where the energy comes from, which community is suffering from the consequences. If you’re planting — if you’re placing a power plant or a data center in a region suffering from water shortage or water scarcity, where communities are struggling, where farmers and rural communities are struggling, when Indigenous peoples are struggling, then you’re imposing additional costs to those people. Normally, they’re not the main beneficiary of the data centers. They’re not the service users. They’re not the investors. And what we are seeing around the world is there is a lot of interest from the private sector in this market, a lot of push for further development, and the placements that are — that are not based on the comprehensive thinking about all these impacts and consequence and what happens in the long run.

In April, we also published a report about the critical minerals and the extraction of critical minerals, showing what is happening in Africa, showing what is happening in South America, where communities are suffering from the pollution. And essentially, we are seeing the reproduction of the oil extraction era. Critical minerals are the fuel of the 21st century. We are seeing a new form of, if you will, like, an imperialism, where we keep extracting those resources. Those people don’t have EVs, don’t even have horses and donkeys, and we move their resources so I can drive my EV in North America and claim that we are contributing to a cleaner future. But in a way, we are saving the future generations in the rich countries of the world at the expense of killing people in some poor communities of the world. This is very dark, but it’s a reality. It doesn’t mean that critical minerals — 

AMY GOODMAN: Very quickly —

KAVEH MADANI: — are bad. It doesn’t mean that —

AMY GOODMAN: Very quickly, what are you calling for, Kaveh Madani?

KAVEH MADANI: Further transparency, further thinking about these things, further taking responsibility along the supply chains. And we are saying there that it’s not only the developers who have to make decisions, but governments also have responsibilities, and us, users, also can make — can be less consumptive. We now have an AI consumerism that we should be worried about. We can use AI less, and more responsibly.