With AI, global energy consumption and GHG emissions are on the rise

+150% in GHG emissions for four companies specialising in AI

Some key figures:

• The report notes an increase of 12% per year in the electricity consumption of data centers, fuelled in particular by the development and deployment of AI (between 2017 and 2023). "This is four times faster than the global growth in electricity consumption", it adds;

• four major companies specialising in AI have seen their GHG emissions from scopes 1 and 2 increase by an average of 150% since 2020 over the period in question;

• The 164 digital companies that reported their electricity consumption accounted for 2.1% of global electricity consumption. Just 10 companies are responsabel for half of this total.

More tech companies set Scope 3 targets

However, progress is being made by these tech companies:

• Nearly half of the companies assessed have committed to achieving carbon neutrality, with 51 of them setting a timeframe shorter than 2050 and 41 projecting to 2050;

• 11% of companies on the panel were running on 100% renewable energy in 2023, an increase of 3 points compared with 2022 (23 companies in 2023 compared with 16 in 2022);

• 55% of companies are publishing targets for indirect emissions linked to their supply chains and the use of their products (scope 3), an increase of 19 points compared to 2022.

Revealing the environmental footprint of AI

In particular, the ITU and the WBA recommend that tech companies:

• publish the full environmental footprint of their AI-related activities;

• Encourage cross-sector collaboration between technology companies, energy producers and environmentalists, alongside industry initiatives to accelerate digital decarbonization;

• continue to accelerate the adoption of renewable energies.

Sophie Bridier