Google's Cloud Servers Gulp 10% of Global Data Center Electricity

Google’s data centers are still consuming much energy despite the firm prioritizing sustainability measures. According to Stocklytics.com, Google’s data centers consumed more than 24 terawatt-hours (TWh) in 2023. This consumption is approximately 10% of the electricity data centers use worldwide, underscoring Google’s expansive digital infrastructure footprint.

The site’s financial analyst, Edith Reads, commented on the analysis:

Google’s heavy investments in AI and machine learning technologies, which are computationally demanding, have led to a significant increase in data center energy consumption.

Stocklytics Financial expert, Edith Reads

Google’s Energy Consumption and Emissions

Google’s data center electricity usage rose significantly by 17% in 2023, mirroring the expanding demand for digital services and the augmented work on AI and cloud computing. Moreover, the increased data center energy usage and supply chain emissions translated into a rise in Google’s total GHG output to 14.3 million tCO2e.  

The aggravated emissions culminated in a 13% year-over-year increase and a 48% ascent compared to their 2019 baseline target, highlighting the difficulties Google struggles with lowering emissions as computing intensity rises and the company expands its technical infrastructure investments to support AI transition. However, that is not to say Google has done nothing to keep its emissions low.

While its current emissions are astoundingly high, Google has made efforts to reduce emissions through fuel routing, reducing almost 2.9 million metric tons of GHG emissions since 2021. This achievement is equivalent to removing roughly 650,000 fuel-based cars off the road for a year.

Google’s Commitment To Sustainability

In line with agreements discussing environmental sustainability, Google is on track to limit its carbon output in the long run. Its work in cloud and AI services was the main driver of its augmented energy consumption last year; however, Google believes that its developments in AI could be the solution the firm needs to maintain sustainability and reduce carbon emissions. The tech giant is on a decarbonization campaign and is set to have all its power grids run on 24/7 carbon-free energy by 2030. About ten grid regions are powered by 90% carbon-free energy, showing  Google’s outstanding progress in grid decarbonization. 

However, using AI and cloud servers translates into escalated AI model training and other computational workloads, which will ultimately augment energy consumption. Thus, Google is working on techniques like quantization and using energy-efficient GPUs to reduce the energy demands required for AI model training. For instance, the firm optimized large-language model training efficiency by close to 40% on Cloud TPU v5e through quantization. The company also wants to integrate Nvidia’s upcoming Blackwell GPU into its Cloud services. Nvidia estimates it will train large models using 75% less power than previous GPU generations for the same task.

Google’s commitment to sustainability will help the firm reduce its carbon footprint and get other companies to emulate similar practices.