• New AI data centers in the US could produce more emissions than an entire nation
• 11 sites have the potential to emit 129 million tons of greenhouse gases per year
• The future sustainability of AI remains unclear

A report has found new AI data centers in the US could have the ability to produce more greenhouse gases than an entire nation, putting the technology under further scrutiny and raising more questions about just how sustainable it is.

An investigation by Wired has revealed that 11 gas-powered AI data centers could generate more greenhouse gases than the whole of Morocco did in 2024, with a potential 129 million tons emitted per year from the sites.

And it's not just one AI company that's being linked to these new data center campuses, with OpenAI, Meta, Microsoft, and xAI all mentioned in the report.

Article continues below

Wired notes that the projects have either been announced, or are already under construction, as AI companies look to bypass the restrictions of the power grid in the US by building their own gas power stations for the sole intention of powering the data centers.

This level of emissions could play a role in the ongoing battle with climate change. The construction of gas-powered stations could impact local communities, and the actions might lead AI users to question whether this is a technology they want to support.

Analysis: AI can be great, but at what cost?

There's no doubt AI can do some impressive things, such as solving a 12-year-old math problem, but it also has its fair share of critics. From hallucinating results and a perceived lack of quality in some circumstances, to well-documented sustainability problems and climate issues.

The Wired investigation highlights that a natural gas project Microsoft is looking to purchase power from "could emit more than 11.5 million tons of greenhouse gases each year". That's more than Jamaica's yearly emissions.

Meanwhile, xAI's (the firm behind Grok) gas turbines in Memphis, Tennessee and Southaven, Mississippi could each emit 6.4 million tons of CO2 equivalents per year.

The list goes on, as AI data center projects in Texas, New Mexico, Ohio, Wisconsin, and more, for the likes of OpenAI and Meta, are detailed in the full report.

It's worth noting the emissions data is from models of power plants constantly running at full capacity, which doesn't happen in reality. Emissions could be two-thirds less than what the models suggest, but that would still be a considerable output of greenhouse gases.

There's no guarantee that all the gas-powered stations mentioned in the investigation will get built, but as demand for AI processing remains strong technology companies will likely look to keep up.

Hope remains though, as some believe AI can be a catalyst for sustainability, but for now the path to cleaner AI is far from set in stone.

Follow TechRadar on Google News and add us as a preferred source to get our expert news, reviews, and opinion in your feeds. Make sure to click the Follow button!

And of course you can also follow TechRadar on TikTok for news, reviews, unboxings in video form, and get regular updates from us on WhatsApp too.

John has been a technology journalist for more than a decade, and over the years has built up a vast knowledge of the tech industry. He’s interviewed CEOs from some of the world’s biggest tech firms, visited their HQs, and appeared on live TV and radio, including Sky News, BBC News, BBC World News, Al Jazeera, LBC, and BBC Radio 4.