How Google’s AI Chip Upgrades Set Sustainability Standards
The global rise of AI has brought increased scrutiny of its environmental impact.
Training large AI models requires significant computing power, with some estimates suggesting individual model training can generate carbon emissions equivalent to five cars over their lifetimes.
This energy consumption has prompted technology companies to examine ways to reduce the environmental footprint of AI computing – and the solutions range from more efficient algorithms to specialised hardware designed specifically for AI calculations.
However, the industry has lacked standardised ways to measure and compare the environmental impact of different approaches to AI computing, which has made it difficult for companies and researchers to assess the effectiveness of various solutions.
Now, the environmental impact of AI computing has decreased through improvements in specialised processor design, according to new research from Google.
Google measures environmental impact of AI chips
Google developed a measurement called Compute Carbon Intensity (CCI) to assess emissions per unit of computation, measured in grams of carbon dioxide per Exa-FLOP - a standard unit for measuring computer processing capability.
The metric aims to increase transparency and drive innovation across the technology industry by providing a standard way to measure and compare the environmental impact of AI processors.
The company tracked improvements from the TPU v4 processor to the current Trillium design and the results show customers using newer TPU generations produce fewer carbon emissions for equivalent AI processing tasks.
Robert Little, Sustainability Strategy Lead for gTech at Google, says: "Their efficiency impacts AI's environmental sustainability. This progress is due to more efficient hardware design, which means fewer carbon emissions for the same AI workload."
Power consumption dominates processor lifecycle analysis
The study revealed that 70% of emissions across a TPU's lifecycle come from operational electricity usage, highlighting the importance of both chip efficiency and clean energy sources.
This finding has led Google to focus on two parallel approaches: improving the energy efficiency of the processors themselves and reducing the carbon intensity of the electricity used to power them.
Meanwhile, manufacturing emissions form the remainder of the environmental impact.
Additionally, the detailed lifecycle assessment helps Google target manufacturing improvements where they will have the most impact - and the company reports it is working with suppliers to reduce the carbon footprint of chip production.
The proportion of total emissions from manufacturing is expected to increase as operational emissions decrease through improved efficiency and cleaner electricity sources.
Industry implications for AI sustainability
The research provides a framework for the wider technology industry to assess and improve the environmental impact of AI systems, as the CCI metric enables companies to track progress and compare different approaches to AI acceleration.
Adam Elman, Director of Sustainability EMEA at Google, says: "This is just the beginning with huge opportunities to continue optimising hardware and software for carbon efficiency."
Robert adds: "These findings highlight the importance of optimising both hardware and software for a sustainable AI future.
“It's important to remember where AI has important implications for reducing emissions and fostering sustainability."
Explore the latest edition of AI Magazine and be part of the conversation at our global conference series, Tech & AI LIVE.
Discover all our upcoming events and secure your tickets today.
AI Magazine is a BizClik brand
