Google admitted that compared to 2019, it produced up to 48 percent more CO2 emissions this year. The reason is the advent of artificial intelligence.
Photo: Google
Google data center.
According to many, artificial intelligence is a revolution at the level of the Internet. The changes that are being introduced are so fundamental that they already have an impact on many otherwise routine activities. In turn, it brings new possibilities for working with data to technological companies. However, it has its own toll that few people were concerned with at the beginning – extreme energy consumption. This is most recently confirmed by Google, which states in its environmental impact report that its CO2 emissions are up to 48 percent higher than in 2019. The reason? The company consumes more and more electricity to run data centers. The exponential increase in the use of artificial intelligence and its integration into services in the company’s portfolio also had a fundamental impact.
A few years ago, Google presented its plan to become a carbon-neutral company by 2030. But in a new report, it admits that with the further integration of artificial intelligence into products, reducing emissions is much more challenging. An increasing proportion of traditional data centers, which have the largest share of electricity consumption, are now used by artificial intelligence applications. To illustrate, a generative AI system such as ChatGPT can consume roughly 33 times more energy than conventional application-specific servers.
The American giant has its servers located all over the world. While in Europe and North America most of them use energy from carbon neutral sources, in the Middle East, Asia and Australia the energy mix is less green. Despite this, Google claims that roughly two-thirds of the generated energy it relies on comes from green sources, such as wind and hydropower.
One of the ways in which Google could reduce the energy consumption of data centers is the utilization of data. A large amount of so-called “black data” is stored on its servers, i.e. files that are not used at all. Such files can be records of system activity, e-mail archives, unused customer data, and the like. Even if the data is not used, the servers on which it is stored must continue to be powered. According to estimates, up to 65 percent of all data that companies store on servers is just such types of data.
Source: vat.pravda.sk
