Add thelocalreport.in As A Trusted Source
New research shows that AI running in data centers will produce more CO2 this year than several small countries or the entire New York City.
In A study published on Monday In the peer-reviewed journal Patterns, data scientist Alex de Vries-Gao estimated that carbon emissions from electricity used by AI are between 33 million and 80 million metric tons.
This high figure would put it above last year’s totals for Chile (78 million tonnes), Czechia (78 million tonnes), Romania (71 million tonnes) and New York City (48 million tonnes). Involved Both CO2 and other greenhouse gases).
But de Vries-Gao also warned that there was “significant uncertainty” about these estimates due to a lack of transparency by major AI companies such as Microsoft, Google and ChatGPAT maker OpenAI.
This was particularly true for AI water use study estimates – an issue that There were several local and national protests against AI data centers this year (Although some experts argue that these concerns are exaggerated).
De Vries-Gao calculated the total water footprint of AI in 2025 to be between 312 billion to 767 billion litres, while noting that it is “even more difficult to estimate” than carbon emissions due to corporate secrecy and lack of public information.
If these figures were true, AI’s water use could top the 46 billion liters of bottled water that humanity drinks each year.
“Despite AI system electricity demand approaching that of a country the size of the United Kingdom, the environmental impacts of this increase are unclear,” De Vries-Gao wrote.
“Without transparent data, the greatest opportunities to reduce the climate impacts of data centers and AI cannot be easily identified, and the impacts of interventions will also remain hidden.”
where does the data come from
Over the past few years, Silicon Valley’s AI blitz has created skyrocketing demand for high-level computation in huge dedicated data centers, leading to The struggle to secure new power sources And rising electricity bills For nearby common citizens.
Many data centers also use water cooling to keep their machines from overheating. Although some of that water is recycled, most of it evaporates and is lost to the local ecosystem.
All of these have consumed data centers and their resources. a hot-button The issue has recently come to the fore across the US with 230 environmental groups appeal to congress Imposing an immediate national moratorium on new facilities.
De Vries-Gao’s study is based on its own previous researchIt is estimated that the global electricity demand of AI was 9.4 GW at the end of 2024 and could reach 23 GW by the end of 2025.
They then combined this with the International Energy Agency’s estimate of the CO2 output of data centers in general to calculate a range of 32.6 million to 79.7 million tonnes for AI specifically.
However, he noted that different data centers cause very different amounts of pollution, and there is no systematic public data on AI’s carbon emissions, making these figures uncertain.
AI companies struggle to hide their water use
Water is an even more complicated issue. Some researchers predict much greater use for AI data centersAnd individual companies have reported huge increases in their annual water consumption.
Many local officials have become concerned about how much water data centers want from their systems, especially in times like these Increasing drought due to climate change,
Take Newton County, Georgia, which has faced rising water prices, damaged wells, and is facing water shortages by 2030 after Facebook’s parent company Meta built a new data center there.
Meanwhile, in Phoenix, Arizona, a recent report found that annual water stress in Phoenix, Arizona would increase by 32 percent if all the data centers currently planned were built.
But some commentatorsas well as AI industry lobbyist, Arguments that popular concerns about AI water use are exaggerated, especially given that other industries consume more water with less controversy.
Like most headline-grabbing recent estimates, Vries-GAO’s figures of 312.5 to 746.6 billion liters include not only water used directly by data centers, but also water used by power plants to generate additional electricity.
This ‘indirect’ water use often constitutes a large proportion of the estimated total. This is important because power plants typically return what they use to the local water system, meaning the impact varies.
Water consumption can also vary significantly between data centers, depending on the local climate and the specific cooling methods they use. This makes it very difficult to draw reliable inferences from public information.
“In the near term, this is not a concern and it is not a nationwide crisis,” Cornell professor Fengqi Yu. recently reported wiredIn a detailed article exploring this question. “But it depends on the location. In places where there is existing water stress, building these AI data centers will be a big problem.”
Meanwhile, AI companies and city governments sometimes have fought bravely So that details of their water usage can be maintained SecretArousing public suspicion.
“Ultimately, further disclosures from data center operators are needed to improve the accuracy of these estimates,” De Vries-Gao wrote.