Study shows AI will triple CO2 emissions in data centers by 2030

According to a study by financial institution Morgan Stanley, the high demand for productive AI is expected to triple greenhouse gas (GHG) emissions by the end of the decade. The analysis found that since the technology has become widespread, Data centers are starting to produce even more CO2.

The study, “Global Data Centers: Sizing and Solving for CO2,” currently assesses global emissions from data centers across all fronts (Scope 1, 2, and 3) and provides forecasts through the end of the decade. The analysis takes into account new construction, growth in existing structures, and energy demand.

According to the research, Server industry on track to emit 2.5 billion tons of greenhouse gas By 2030, this number is three times higher than what would be predicted if generative AI were never invented.

Of these 2.5 billion tons of CO2, 60% should be a result of the higher energy demand of each buildingThe remainder is added to other processes, including the manufacturing of new components and the preparation of each structure.

Morgan Stanley estimates that carbon emissions from building data centers will rise from 200 million tons this year to 600 million tons by 2030.

Morgan Stanley’s research seeks to highlight the importance of developing more energy efficient solutions or reversing the situation of unlimited consumption with solutions that reduce the impact of new technologies. The institution cites the possible growth of markets such as carbon capture programs (CCUS).

It is also expected that the giants of the sector (Google, Meta, Microsoft and Apple) will place greater demands on carbon dioxide removal. In total, around 25 megatons of CO2 equivalent by 2030.

Reforestation projects and other initiatives are also expected to grow significantly to meet giants’ commitment to achieve carbon neutrality – Microsoft, for example, wants to be “carbon negative” by 2030.