Google is relying more and more on energy-intensive data centres to power its new artificial intelligence products, which puts its objective of lowering its carbon footprint in peril. The IT giant disclosed on Tuesday that during the previous five years, its greenhouse gas emissions have increased by 48%.
According to Google, the main reasons for the increase were supply chain emissions and data centre electricity consumption. Additionally, it disclosed in its yearly environmental report that its emissions reached 14.3 million metric tons in 2023, an increase of 13% over the year before.
The tech giant, which has made significant investments in artificial intelligence, stated that achieving its “extremely ambitious” target of net zero emissions by 2030 “won’t be easy.” It stated that “the uncertainty around the future environmental impact of AI, which is complex and difficult to predict,” was one of the “significant uncertainties” around achieving the goal.Since 2019, when Google set the baseline for achieving net zero emissions—a target that necessitates the firm reducing as much CO2 as it emits—its emissions have increased by about 50%.
According to estimates from the International Energy Agency, the overall electricity consumption of data centers might double from 2022 levels to 1,000TWh (terawatt hours) in 2026, or roughly the same as the amount of electricity used by Japan. According to estimates made by research firm Semi Analysis, artificial intelligence would cause data centers to use 4.5% of the world’s energy production by 2030.
Data centers are essential for the training and maintenance of AI models such as Open Ai’s GPT-4, which powers the ChatGPT chatbot, and Google’s Gemini. Microsoft said this year that its “moon-shot” goal of becoming carbon negative by 2030 was in jeopardy due to energy consumption associated with its data centers. Microsoft President Brad Smith acknowledged in May that the company’s AI strategy is to blame for “the moon moving.” Bill Gates, a co-founder of Microsoft, stated last week that artificial intelligence (AI) might aid in addressing the climate crisis since large corporations are “seriously willing” to pay more for clean energy sources so as “to say that they’re using green energy. Large tech firms have increased their purchases of renewable energy significantly in an effort to reach their climate targets.
But promises to cut CO2 emissions are now up against promises to heavily invest in AI products, which need a lot of energy for training and deploying in data centers, not to mention the carbon emissions from the production and shipping of the chips and computer servers needed for that process. Another environmental concern contributing to the AI development is water usage. According to one research, by 2027, AI could account for up to 6.6 billion cubic meters of water use, or roughly two-thirds of England’s yearly consumption.