You turn to us for voices you won't hear anywhere else.

Sign up for Democracy Now!'s Daily Digest to get our latest headlines and stories delivered to your inbox every day.

Massive Energy Needs of AI & Cryptocurrencies Lead Amazon, Google & Microsoft to Embrace Nuclear Power

Listen
Media Options
Listen

Tech companies are turning to nuclear to fulfill the skyrocketing energy needs of artificial intelligence, with major corporations like Amazon, Google and Microsoft announcing plans to invest in nuclear power. But the speed at which energy needs are growing may not align with the construction or revitalization of nuclear infrastructure, says Alex de Vries, who researches the unintended consequences of AI and cryptocurrencies. There may be a “mismatch between the needs of tech companies today” and the future, while nuclear power continues to carry the same safety risks that led to its phasing out in the first place.

Related Story

StoryNov 19, 2024Nuclear Revival? Diné Activist Warns Against New Uranium Mining as AI, Bitcoin Energy Needs Soar
Transcript
This is a rush transcript. Copy may not be in its final form.

AMY GOODMAN: This is Democracy Now!, democracynow.org, The War and Peace Report. I’m Amy Goodman. We’re broadcasting here at the U.N. climate summit in Baku, Azerbaijan.

We turn now to look at the controversial push here at the U.N. climate summit to expand nuclear power production across the globe. Supporters of nuclear energy view it as a carbon-free energy source that’s needed to combat the climate crisis. But critics warn nuclear power is too expensive and risky.

Last year at the U.N. climate summit in Dubai, more than two dozen countries, including the United States, pledged to triple nuclear energy capacity. In a statement, the countries said expanding nuclear energy is needed for, quote, “achieving global net-zero greenhouse gas emissions / carbon neutrality by or around mid-century and in keeping a 1.5°C limit on temperature rise.” Six more countries joined the pledge last week. The Biden administration recently unveiled a plan to create an additional 200 gigawatts of nuclear energy capacity by midcentury.

The incoming Trump administration is expected to continue the push to expand nuclear energy. Trump’s pick for energy secretary, the fracking magnate, CEO Chris Wright, sits on the board of the nuclear power startup Oklo.

Meanwhile, private tech companies are also investing in nuclear energy. Google has announced plans to build seven small nuclear reactors in the U.S. to help power the company’s data centers and artificial intelligence systems. Microsoft has said it would fund the reopening of the Three Mile Island nuclear power plant, the site of the worst nuclear disaster in U.S. history. And Amazon’s cloud computing subsidiary has signed a deal with Dominion Energy to develop small nuclear reactors.

We’re joined right now by three guests. Alex de Vries, we’re going to go to first. He’s joining us from Amsterdam, a Ph.D. candidate who studies the sustainability of emerging technologies at the VU Amsterdam University. He’s the founder of Digiconomist, a research company dedicated to exposing the unintended consequences of artificial intelligence and cryptocurrencies.

Alex, we’re going to begin with you. Can you explain why there is such an enormous need for energy when it comes to artificial intelligence, AI data centers?

ALEX DE VRIES: Yeah, of course. Well, the thing with generative AI and, well, huge applications like ChatGPT is that they’re really using massive models in the background. And one typical dynamic is that the bigger you make those models, the better they perform, which is what everyone wants. But the downside of that is that bigger models also means more computational resources are required, and thus energy in the end. Where we had multimillion-parameter models just a few years ago, now you’re talking about models with billions or even trillions of parameters, and they just keep on growing, because everyone wants to offer the best user experience. But that costs a lot of resources.

AMY GOODMAN: So, talk about — I mean, I don’t know even how much people understand. I mean, in 2016, you found one bitcoin transaction consumes as much energy as the average American household does in a day. Why does AI, why do cryptocurrencies — we talked about Amazon, we talked about Google, I mean, also Elon Musk with cryptocurrencies, with bitcoin. Why do they require so much electricity?

ALEX DE VRIES: Yeah, they actually both contain economic incentives to keep spending more on energy. In the case of cryptocurrency mining, there is a reward available for those who would participate in the mining process. And ultimately, everyone is incentivized to just, you know, keep putting in more computational power, as long as it’s profitable to do so. So, the more money is available as a reward, which has been growing over the past few years. As the bitcoin price kept going up in value, the more people put in energy-hungry equipment to participate in that process. And with AI, it has very similar incentives in the sense that everyone wants to have the biggest, best and most energy-hungry model, simply because the biggest models offer the best user experience. And that, in turn, will give you the biggest market share, and thus the highest revenues. So, both of these technologies contain different but still economic incentives to keep using more resources to make these applications run.

AMY GOODMAN: So, why is bitcoin responsible for more than 95% of the carbon footprint of all cryptocurrencies?

ALEX DE VRIES: Yeah, that’s because bitcoin uses the very specific mining process where, in order to create a new block for the underlying blockchain, a new block of transactions, everyone in the bitcoin network has to participate in what I like to describe as a massive game of guess the number, whereas the whole network is making an insane amount of calculations every second of the day, 600 quintillion attempts — it’s like 600 with 18 zeros — every second of the day, nonstop. One participant guesses correctly every 10 minutes and then gets to create the next block for the blockchain, and they get a reward for that. These miners are earning up to or more than $10 billion a year from this process. But, of course, they are also spending a huge amount of money on the energy to participate in this process, and nowadays actually more than entire countries, like Sweden and Argentina, just for mining bitcoin. But other cryptocurrencies, they use other types of algorithms, so those algorithms are not dependent on computational effort. This is why bitcoin is kind of the only big cryptocurrency left standing today which still uses the extremely energy-intensive mining.

AMY GOODMAN: So, I wanted to ask you about the organizations and the people who are pushing for nuclear energy to fulfill the demands of AI and cryptocurrency. If you can talk about some of these companies? We know Elon Musk, for example, is pro-nuclear.

ALEX DE VRIES: Yes, of course. Well, there’s actually two avenues here. First of all, what we see is that a lot of these tech companies are looking at existing nuclear energy sources as a way to power their AI operations today. One obvious advantage of nuclear power is that it’s zero-carbon energy. But what they’re trying to do is they’re trying to get this power from nuclear reactors that have previously been shut down. They’re trying to revive these nuclear reactors. And this actually carries a very interesting safety dimension, because if you look at what goes on, is that how these nuclear reactors — they get shut down at a certain point in time. They start to decay, because the moment they shut down, they no longer have to meet safety regulations. And now these big tech companies show up and try to revive these facilities, which are already outdated and haven’t been safe for quite some time, just to power their current AI operations, because they want to have zero-carbon energy. It carries a lot of safety risks.

And then, the other avenue is, going forward, we also see a lot of talk about expanding the nuclear — existing nuclear power supply to fuel future AI energy demand. But then, the thing is, there’s actually quite a bit of a mismatch between the needs of the tech companies today and, well, the ability to deliver that power to them, because if you are building out new nuclear infrastructure, those are long-term projects that can take up to a decade to complete, while we’re in the middle of an AI hype today. And no one is guaranteeing that there will still be this much demand for AI 10 years from now, when your nuclear reactor is finished.

AMY GOODMAN: We’re speaking to you in Amsterdam. The Netherlands had decided to phase out nuclear power but then reversed its decision. There are two new nuclear plants being built there, where you are?

ALEX DE VRIES: Well, yeah, you can see that nuclear power is indeed experiencing a bit of a revival. I guess everyone is realizing that this type of power is zero carbon, and that’s what we need if we’re going to go green. Obviously, it’s not as safe as wind or solar power, but, then again, nuclear reactors nowadays, with the latest technological standards, are still very safe. But it’s just that constructing them is relatively expensive compared to expanding the wind and solar power supplies. So, whether this is the best investment still remains a question.

The original content of this program is licensed under a Creative Commons Attribution-Noncommercial-No Derivative Works 3.0 United States License. Please attribute legal copies of this work to democracynow.org. Some of the work(s) that this program incorporates, however, may be separately licensed. For further information or additional permissions, contact us.

Next story from this daily show

Nuclear Revival? Diné Activist Warns Against New Uranium Mining as AI, Bitcoin Energy Needs Soar

Non-commercial news needs your support

We rely on contributions from our viewers and listeners to do our work.
Please do your part today.
Make a donation
Top