Google’s “ambitious” goal to achieve net-zero emissions by 2030 faces a significant setback as the tech giant’s greenhouse gas emissions have surged by 48% since 2019, according to its latest environmental report released on Tuesday.
The primary driver behind this substantial increase is the rapid integration of artificial intelligence into Google’s core products, which has led to a sharp rise in energy consumption and associated emissions.
Google’s Path to Net-Zero
Net-zero refers to a state in which the greenhouse gases going into the atmosphere are balanced. While an increasing number of companies, including Google, have committed to achieving net-zero emissions in their operations, the advancements in AI technology are making these goals somewhat complicated to achieve.
A report from the International Energy Agency roughly calculates that the total electricity consumption of AI data centres could double from 2022 levels to 1,000 TWh (terawatt hours) in 2026, which is equivalent to the electricity consumption of Japan. Meanwhile, a separate study by research firm SemiAnalysis has revealed that artificial intelligence will cause data centres to consume 4.5% of global energy generation by 2030.
The resource-intensive nature of AI technologies poses risks to companies’ efforts to combat climate change. Despite these concerns, Google remains steadfast in its dedication to achieving sustainability.
“We’re committed to responsibly managing the environmental impact of AI by deploying three major strategies: model optimisation, efficient infrastructure, and emissions reductions,” Google stated in the report.
Model Optimisation
The company has identified and tested practices that, when used together, can reduce the energy required to train an AI model by up to 100 times and associated emissions by up to 1,000 times. These techniques are currently employed at Google.
Google has also taken steps towards accelerating AI model training through quantization, which has boosted the efficiency of large language model training by 39% on Cloud TPU v5e. Additionally, the company’s Go Green Software guide assists developers in reducing their digital footprints.
In terms of deployment and usage, Google’s Gemini 1.5 Pro model delivers notable improvements with quality that is as good as the Gemini 1.0 Ultra, all while using fewer computational resources.
Efficient Infrastructure
Google is in the works to continuously improve the energy efficiency of its AI hardware. In fact, the company’s fourth-generation Tensor Processing Unit (TPU v4) has obtained a 2.7x improvement in its performance per watt compared to TPU v3.
As stated in the report, Google targets offering Nvidia’s next-generation Blackwell GPU to its cloud customers, which Nvidia estimates will train large models using 75% less power than older GPUs for the same task. Moreover, Google’s new Axion Processors are up to 60% more energy-efficient than the current-generation x86-based instances.
Emissions Reductions
In 2023, Google purchased more than 25 TWh of renewable electricity, including from PPAs, on-site renewable energy generation, and grid renewable energy. The tech giant has also deployed innovations, which include advanced geothermal, carbon-intelligent computing, and demand response capabilities. The company also partnered with Microsoft and Nucor to leverage its demand aggregation and procurement models for advanced clean electricity technologies.
Aside from the key major strategies, Google emphasised that harnessing the potential of AI for climate action requires collective efforts. Policymakers, in particular, can help by encouraging data sharing, ensuring affordable technology access, building awareness, and supporting the creation and expansion of AI and climate-related upskilling programmes for corporations. They can also accelerate the deployment of AI for climate change by defining public and private sector priorities, delivering on public sector use cases, and encouraging private sector action. Lastly, policymakers can promote the environmentally and socially responsible deployment of AI.
Achieving a balance between innovation and sustainability will require firm governance frameworks, ethical considerations, and a joint commitment to ensure artificial intelligence will benefit all.