This post is also available in:
עברית (Hebrew)
Google recently announced that its climate emissions had risen by 48 percent since 2019 because of artificial intelligence, while many experts are pointing at the amount of energy the technology is sucking up and its impact on the environment.
In order to look into this growing issue, we must understand how AI uses electricity: Before we even come to using AI tools, the development and training stages require a huge amount of computer power, requiring lots of electricity that generates lots of heat, which needs to be cooled using even more electricity. Then when the tools are operational, whenever we put a request into an AI engine, it is sent to a data center where it is processed using huge amounts of energy.
Furthermore, now that the new exciting trend is using AI tech, experts are worried that this increased (and sometimes unnecessary) use will cause electricity usage to spike since AI services require more power than non-AI ones.
A clear and worrying example shows that each request made to ChatGPT uses 10 times the power of a single Google search, meaning that if Google switched all search queries to AI it would inflate the company’s electricity usage to unimaginable heights.
According to Techxplore, before the era of AI, data centers took about one percent of global electricity demand. In 2022, data centers, cryptocurrencies, and AI combined used almost two percent of total global electricity demand, and it is estimated to double by 2026, which would be equal to the electricity used by the entirety of Japan.
It is important to note that the big players in AI and data centers like Amazon, Google, and Microsoft have reportedly been trying to reduce their carbon footprints by buying vast amounts of renewable energy. An Amazon official told AFP that the company’s data center division AWS was “the largest purchaser of renewable energy in the world today.” Granted, while AWS is committed to being a net-zero carbon company by 2040, Google and Microsoft have pledged to reach that goal by 2030.
However, simply building new data centers and ramping up usage in existing centers is not going to help with green energy targets. Indeed, Google and Microsoft both said in statements that their greenhouse gas emissions have been rising in the last few years (a 48 percent rise from 2019 by Google and a 30 percent increase from 2020 by Microsoft) – both blaming AI.
