Tackling AI & Data Centre Emissions and Reporting Gaps
The environmental efficiency of data centres continues to hit headlines, especially concerning their energy usage and emissions footprint.
The International Energy Agency (IEA) predicts a significant surge in energy demand due to AI advancements, positioning data centres alongside AI as critical elements of future energy consumption.
Data centres and transmission networks each reportedly use up between 1% and 1.5% of global electricity, highlighting the pressing need for improved efficiency amid growing AI applications.
However, varying reports suggest a gap between claims of enhanced efficiency and the often higher actual emissions., with the disparity spotlighting the continued environmental burden of AI expansion on data centres.
Are headline claims a reality?
The collective energy consumption of global data centres annually exceeds 460TWh, equivalent to Japan's entire energy usage.
Projections indicate this could surge towards 1,000TWh in coming years.
The IEA widely estimates that data centres contribute roughly 3% of global carbon emissions, with their environmental similar to that of the aviation sector.
Amid ambitious corporate narratives about reducing emissions, Guardian analysis suggests data centre emissions are possibly 662% higher than reported, especially as AI demand amplifies energy consumption.
Analysis from 2020 to 2022 indicates big tech companies like Google, Microsoft, Meta and Apple reported emissions approximately 7.62 times higher than officially reported.
This discrepancy emphasises the hidden environmental costs associated with powering the digital revolution.
“This is a fast-moving space and these numbers come with uncertainty,” Microsoft’s Senior Director of Sustainability, Dr Amy Luers, says.
“Anchoring discussions in trusted sources like the IEA helps cut through the noise and support more informed dialogue.”
However, Amy points out that this should not mean stats from bodies like the IEA should be the only form of truth.
She adds: “But using estimates from the IEA and similar trusted organisations as benchmarks — especially when analysts and media use them to understand and explain why a figure is higher or lower — can help strengthen the dialogue.
“Let’s keep working to cite data, state assumptions and benchmark wherever possible.”
The major emissions contributors
Identifying a single largest emitter is complex due to differences in reporting standards, data availability and the challenge of accounting for both direct and indirect emissions.
Nonetheless, Amazon Web Services (AWS), as the world's largest cloud provider, is generally recognised as the most significant emitter among data centres.
Microsoft, Google, Meta and Apple also rank as substantial emitters, though some critics argue that their reported emissions often substantially underestimate the true figures.
Amazon does not publicly disclose raw emissions data solely for AWS data centres in its sustainability reports. However, the 2024 report reveals that Amazon’s total carbon emissions increased from 64.38 million MTCO₂e in 2023 to 68.25 million MTCO₂e in 2024.
Despite this rise, the company’s carbon intensity improved, as emissions per dollar of gross merchandise sales dropped from 75.6g CO₂e to 72.6g CO₂e, indicating enhanced operational efficiency amid growth.
AWS data centres are recognised for strong operational efficiency, boasting a global Power Usage Effectiveness (PUE) of 1.15 in 2024, considerably better than the public cloud industry average of 1.25.
This reflects AWS’ ongoing focus on improving energy utilisation across its infrastructure.
All five major tech firms have committed to carbon-neutrality goals, predominantly targeting 2030 as the timeframe to achieve it by.
Google, for instance, aspires to run its data centres and offices on 24/7 carbon-free energy by the end of the decade, and Microsoft aims to be carbon negative by 2030, pledging to remove all carbon emissions produced since its founding.
Google has evolved its approach over time, ending its large-scale purchases of cheap carbon offsets and thus retreating from its earlier claim of carbon neutrality since 2007. It now targets net-zero carbon emissions by 2030.
According to Google’s 2024 Environmental Report, the company’s total greenhouse gas emissions rose by 13% year-on-year in 2023, driven notably by increased AI and data centre energy demands.
In the introduction to the report, Google CSO Kate Brandt and Benedict Gomes, SVP, Learning & Sustainability, say: ”In spite of the progress we are making, we face significant challenges that we’re actively working through.
“In 2023, our total GHG emissions increased 13% year-over-year, primarily driven by increased data centre energy consumption and supply chain emissions.”
“A sustainable future requires systems-level change, strong government policies and new technologies. We’re committed to collaboration and playing our part, every step of the way.”
At Amazon, the company surpassed its target to match 100% of the electricity consumed by its global operations with renewable energy by 2030 — a full seven years ahead of schedule.
Alongside this milestone, Amazon achieved a 3% reduction in absolute carbon emissions and a 13% decrease in carbon intensity.
While AI presents some challenges to Amazon’s sustainability credentials, the company is harnessing AI to optimise its solar and energy storage farms that power data centres.
AWS infrastructure is now up to 4.1 times more energy efficient than on-premises setups and has the potential to reduce workloads’ carbon footprints by as much as 99%.
To maintain this momentum, Amazon continues to expand its portfolio of more than 600 renewable energy projects worldwide.
It has a renewable energy footprint in 28 countries and has held the position of the largest corporate purchaser of renewable energy globally for five consecutive years.
Looking ahead, AWS also aims to be water positive by 2030, committing to return more water than it consumes.
The future of data centre and AI emissions
Projections by Morgan Stanley indicate AI-optimised centres could quadruple power usage, with global emissions from such sites potentially tripling to 600 million metric tons by 2030.
New data centre construction intended to accommodate AI might further elevate carbon footprints, complicating emission reductions.
For tech leaders scaling infrastructure, these sustainability pledges might face strategic challenges, heightening dependence on renewable investment, carbon capture innovation and improved efficiencies.
To achieve net-zero goals while advancing AI-enabled digital growth, significant adoption of renewable energy and policy adjustments will be crucial.
As Brad Smith, Microsoft’s Vice Chair and President, said in its sustainability report: “We believe data transparency is critical to driving accountability and progress on climate.”
