On February 11th, 2025 AI Now Co-Executive Director Sarah Myers West delivered the plenary keynote at the Forum on AI and Sustainability at the AI Action Summit in Paris. Watch the full recording here or read her remarks below:
Thank you for having me today. With Amba Kak I lead the AI Now Institute, a think tank that exists to challenge the current trajectory for AI and bring it in line with the public interest. We do this work in collaboration with allies from many corners of society from tech policy to economic justice to labor groups, and I wish more of them were in this room today because we all have stakes in this issue.
We are approaching a critical threshold in our computational futures and we’re going to need all the help we can get. The AI industry and governments are all in on full-scale development of AI infrastructures, to support conducting ever larger––and implicitly more powerful––AI model development. But there are many reasons to be skeptical of AI boosterism; while it’s very clear that promoting public funding for AI carries many benefits to companies, the benefits to the broader public are much more sketchy and hypothetical. With the push to build AI at ever larger scale comes many pathologies, from the privacy and security harms to the proliferation of false content to what brings us all here today: harms to the climate. If there’s one key takeaway I’d like to leave you all with at the start of the day, it’s this: the current trajectory of AI development at all costs is not only not necessary, as was exemplified by the recent release of the more efficient model Deepseek. We must depart from it urgently or we will enact damage that we will be unable to roll back.
New research by Beyond Fossil Fuels released this week shows the strain that data center growth will put on Europe alone: electricity demand by data centers in Europe could rise by up to 160% by 2030, more than Spain’s total electricity consumption in 2022. Emissions could reach 121 million tons of carbon dioxide, wiping out over half of Germany’s planned emission reductions over the same period. This research should be a wake-up call for decision-makers that the concept of sustainable and limitless growth for data centers is a myth.
Now, we’re hearing a lot about ways to make AI more efficient, plans to run data centers on clean energy, and that AI may even offer benefits for sustainability. That makes this message sound quite stark. But it’s reflective of movements happening outside of this room, where people in far-flung parts of the globe from Indiana to Chile to Taiwan to the Netherlands are contending with the cold, hard reality that there is no version of the current AI boom that will lead to a sustainable future. And instead of meaningfully grappling with this reality, governments are responding by criminalizing protest. But it’s their water, their air, their land that is being damaged by this buildout. And while things may feel glowy within the confines of this Summit, there is a confrontation brewing that we can’t look away from.
Because the AI industry’s insatiable energy appetite is creating an unprecedented strain on power supplies. Rising demand is pushing power infrastructure in parts of the world to its limits and prolonging our dependency on fossil fuels, while also exhausting other critical resources like water, land and raw materials. This is a public health issue resulting in increased pollution and even premature death, which experts estimate will cost $20 billion per year in the United States alone.
False promises and false solutions abound in this debate. Major tech companies like Google and Microsoft are now questioning if they can hit their own climate and energy targets, while Amazon and Meta engage in efforts to enable them to continue to burn fossil fuels while claiming to be 100% renewable. Meanwhile, nuclear energy is being touted as a solution, but it is largely a dangerous distraction. The long timeline of Small Modular Reactors and safety risks of using decommissioned nuclear plants further undermine the credibility of this approach. On top of this, AI tools are sold to the oil and gas industry, enabling millions of tonnes more in carbon emissions at a time when the scientific consensus clearly calls for a phase out of fossil fuels. AI can never be a ‘climate solution’ if it runs on fossil fuels and is used to scale up the extraction of oil and gas.
This is why last week the Green Screen Coalition, Green Web Foundation and over a hundred other organizations released a statement with a powerful call: Development of AI infrastructure must be restrained within the limits of what the planet can sustain. This means the AI industry must immediately phase out using fossil fuels. Companies must take responsibility for the emissions produced within their supply chains, and both measure and make transparent environmental impacts across the AI lifecycle.We don’t need to tie ourselves to this unsustainable vision for our future: there is nothing about this technology that is inevitable. It has taken many turns over the course of an almost 70-year history and on my more optimistic days, I do this work in the hope that a new approach is around the corner. Critique is a belief that there is another world that is possible. To the policymakers in the room, we have tremendous scope to shape the trajectory for AI to ensure it serves the needs of the public, by mandating an end to the growth at all costs approach and turning to more sustainable paths. The market will not get us there alone. I appeal to all of you today to do so.
