AI boom drives 150% surge in Big Tech emissions
AI emissions could exceed some countries’ in the near future
AI is scaling the tech industry’s emissions to dangerous levels, says a new report.
A new report from the United Nations' International Telecommunication Union (ITU) has revealed a 150% average increase in indirect carbon emissions from four leading AI-focused tech companies – Amazon, Microsoft, Alphabet and Meta - between 2020 and 2023.
The spike is largely attributed to the high energy demands of AI systems and datacentres, which account for a rising portion of global electricity consumption.
AI and machine learning contributed 0.2% of global emissions in 2021, which even then was about 10% of the entire tech sector. It has accelerated from there.
Among the worst performers, Amazon's operational emissions increased by 182% over the 2020-2023 period, followed by Microsoft (155%), Meta (145%) and Alphabet (138%).
These include both direct emissions from company operations and those associated with their purchased electricity use (indirect).
The ITU says unchecked AI expansion could result in annual emissions reaching 102.6 million tonnes of CO2 equivalent - comparable to the yearly emissions of some mid-sized industrialised nations.
Currently, no international standards or regulations require companies to report emissions specifically attributable to AI systems or their energy use, leaving significant blind spots in accountability.
This is an issue we have grappled with in our annual Cloud Sustainability metrics, which this year we expanded to include vendors outside just US hyperscalers.
"Currently, there are no standards or legislative requirements for companies to disclose their AI emissions or energy consumption, which makes understanding the impact of AI on company-level energy use less straightforward," the ITU report says.
"However, data from company reports show an increasing trend in operational emissions for companies with a high level of AI adoption."
The report is based on disclosures from 164 of the world's 200 largest digital companies and includes a detailed assessment of energy consumption.
In 2024, tech companies consumed an estimated 581 terawatt-hours (TWh) of electricity – about 2.1% of global power demand – with just 10 companies accounting for more than half (51.9%) of that total.
These top energy consumers include China Mobile, Amazon, Samsung Electronics, China Telecom, Alphabet, Microsoft, TSMC, China Unicom, SK Hynix, and Meta.
Datacentres – the physical backbone of AI and cloud services – are identified as a central pressure point.
According to International Energy Agency (IEA) figures cited in the report, datacentre electricity consumption has increased about 12% annually since 2017, reaching 415 TWh in 2023.
If this growth continues, datacentres could use 945 TWh by 2030, surpassing the total annual electricity consumption of Japan.
"The rapid growth of artificial intelligence is driving a sharp rise in global electricity demand, with electricity use by data centres increasing four times faster than the overall rise in electricity consumption," the report found.
In total, the 166 companies that publicly disclosed their emissions were responsible for 297 million tonnes of CO2 equivalent in 2023 – equivalent to the combined emissions of Argentina, Bolivia and Chile.
In light of the ITU's findings on the digital revolution's environmental cost, there are heightened demands for governments and industry leaders to establish clear sustainability standards and transparent emissions reporting frameworks for new technologies.
The report is expected to intensify pressure on major digital companies and regulators to prioritise sustainable AI development, and to bring greater transparency to the environmental costs of digital infrastructure.