Google’s Emissions Jumped 51% As AI Data Centres Push Up Demands Hindering Green Energy Goals 

Google’s year-on-year electricity consumption surged 27%, indicating its inability to decarbonise as its energy needs are soaring. 

Countries and private companies alike are in a race to develop their artificial intelligence (AI) capabilities, and every day, news of newer breakthroughs hits the headlines. However, the environmental costs are rarely discussed. Google recorded a 51% increase in carbon emissions as the company’s AI goals are hindering its green ambitions. 

When weighed against technological and financial targets, environmental pledges are often compromised. However, with the climate emergency at our doorstep, ignoring climate goals is simply unaffordable. While companies and governments are pledging millions of dollars to combat climate change challenges, such progress seems minuscule and is offset by other unsustainable heavy industrial or technological developments. 

Similarly, Google also invested in renewable energy and carbon removal technology. However, it has failed to reduce its scope 3 emissions, which are far down the supply chain and are a result of increasing data centres, which are needed to power AI. Google’s year-on-year electricity consumption surged 27%, indicating its inability to decarbonise as its energy needs are soaring. 

According to estimates by the International Energy Agency, data centres’ total electricity consumption could double from 2022 to 1,000TWh (terawatt hours) next year. The Guardian reported that research firm SemiAnalysis forecasts that to meet AI demands, data centres’ energy usage could comprise 4.5% of the total global energy generated by 2030. SemiAnalysis has also cautioned that unprecedented AI growth could lead to ‘non-linear growth’ in energy demands, making it difficult to assess future energy demand and consumption patterns. 

However, Google has complained that its electricity consumption has surged as little progress has occurred in new forms of low-carbon electricity generation. Companies do have options to decarbonise if they adopt Small Modular Reactors (SMRs) are miniature nuclear plants and are easy to build and install to get on the grid. Areas with numerous data centres could have more than one SMR, which can significantly reduce the carbon footprint from the energy used by these data centres. 

However, there are delays in its adoption as building SMRs is still in the nascent stage, expensive and poorly incentivised under the existing regulatory framework. Google has acknowledged that it remains committed to adopting renewable energy sources, but the delayed deployment of these SMR and geothermal technologies is hampering the company’s attempts to reduce its carbon dependency by the end of this decade. 

Scope 3 has been identified as the main challenge, given that Google’s ambition-based emissions were 11.5 million tons of carbon dioxide-equivalent gases, up 11% from the year before and 51% from the base year of 2019. Supply chain emissions were the main driver of this increase, accounting for the largest portion of the 22% increase in scope 3 emissions in 2024. 

Google wants to power its systems with clean energy and has signed over 170 agreements to date from 2010 to purchase over 22 gigawatts (GW) of clean energy. Last year, 25 of these agreements came online to add 2.5GW of clean energy to its operations. While emissions may have jumped, 2024 also marked a record for Google, as it signed contracts for 8GW in clean energy. 

It also fulfilled another environmental target last year, having gone plastic-free. Google announced last week that packaging for newly launched products manufactured in 2024 was 100% devoid of plastic. Having hoped to achieve this by the end of this year, the company has met its own deadline a whole year earlier. 

Google also reported that AI can help reduce more emissions than it produces. The company is working towards building AI models which can assess and predict energy consumption patterns and help reduce wastage, and also map the placement of solar panels to generate the maximum amount of electricity. Google is working with various partners to reduce 1GT (gigaton) of their carbon-equivalent emissions annually by 2030 with the help of its AI products. 

Given Google’s focus towards building products which could accurately predict energy consumption patterns, the company’s increased emissions almost seem like a necessary evil in the short run as it tries to come up with long-term AI-based solutions to reduce emissions.