Artificial Intelligence Increases Amazon's Environmental Footprint
Not just Google. The expansion of infrastructure dedicated to artificial intelligence is significantly impacting Amazon's environmental commitment, according to a sustainability report recently published by the company and echoed by an analysis from Axios. The document, which snapshots the year 2025, highlights a 34% increase in total electricity consumption compared to the previous year, a leap that translates into a 16% rise in greenhouse gas emissions: the highest value ever recorded by the company.
The environmental balance is affected not only by electricity. The report indicates that Amazon's data centers used about 2.5 billion gallons of water over the year. The document also reports a 10% growth in emissions classified as 'indirect', a category that primarily includes production and supply chains. Compared to 2019, a reference year adopted by many companies for their climate goals, the increase in this area exceeds 21%.
A detail that did not go unnoticed concerns the placement of the most critical data within the document: the table summarizing these numbers is found on page 46 of a report that totals 51 pages, towards the end of the text.
Alongside the publication of the data, Amazon has highlighted some initiatives presented as priorities for the fight against climate change, including a specific commitment against so-called "super pollutants," substances that the report itself defines as more harmful than common carbon dioxide, in equivalent quantities. This initiative, signed along with other companies including Google, Figma, Salesforce, and JPMorgan starting in March, provides for a collective commitment of $100 million to reduce these types of emissions. However, the report does not specify what portion of the funding is actually attributable to Amazon relative to the total.
The overall picture emerging from the document thus highlights a contradiction between the company's stated goals - which in 2024 confirmed its commitment to achieving carbon neutrality by 2040 - and the actual trend of energy consumption and emissions, which are increasing precisely in the year of greatest expansion of artificial intelligence-related infrastructure. It remains to be seen whether the announced initiatives will be sufficient to reverse a trend that, numbers in hand, is proceeding in the opposite direction to the declared objectives.