Editor’s note
Welcome to another edition of the Jisc Digital Sustainability Newsletter.
Whilst researching and compiling the stories and papers for this month’s edition, I noticed that many of the resources highlight the same uncomfortable truth. There seems to be a growing sentiment that we can no longer ignore the gap between corporate sustainability claims and their real-world effectiveness. Many of the metrics, frameworks and certification schemes we rely on to assess environmental performance often fail to capture what actually matters. We can see this in LEED scores that don’t reflect real emissions savings, renewable energy accounting that obscures actual grid dependence, and the fact that AI sustainability research is so dominated by technofix thinking that it rarely asks whether a given system should be built at all.
There are, however, some more positive, constructive threads too. The development of a new framework for measuring software emissions across its full life cycle, a tool for mapping local carbon intensity rather than relying on national averages, and a paper making the case for why smartphone settings design matters more than we think. For me these could potentially signal a shift towards a more granular and honest accounting of digital environmental impact. My only hope is that initiatives such as these can move beyond academic theory and be integrated into government policy and industry standards.
As ever, I hope you find this edition useful. Do get in touch if you have stories, papers or resources to share, and subscribe for future editions of the Digital Sustainability Newsletter via the DIGITAL-SUSTAINABILITY JiscMail list if you haven’t already.
– Cal Innes, Digital Sustainability Specialist, Jisc (cal.innes@jisc.ac.uk)
Digital sustainability news
Here’s a quick roundup of this month’s biggest digital sustainability news headlines:
Big Tech lobbied EU to keep data centre environmental data secret, investigation reveals
Microsoft and industry lobby group DigitalEurope successfully pushed to classify individual data centre environmental metrics as commercially confidential under EU law, blocking public access to information on energy and water use, a move legal experts warn may violate international transparency obligations.
Iran conflict exposes green energy supply chain’s dependence on vulnerable shipping routes
The conflict with Iran has disrupted not just oil and gas markets but also the maritime routes carrying the critical minerals needed for electric vehicles, wind turbines and solar panels, revealing that the clean energy transition faces many of the same geopolitical vulnerabilities as the fossil fuel system it aims to replace.
England to require solar panels and heat pumps in all new homes from 2028
The UK government has announced that all new homes in England will be required to have solar panels and heat pumps installed under updated planning rules, with plug-in balcony solar panels also set to become available in UK supermarkets for the first time.
Amazon releases new cloud emissions dashboard
Amazon has launched a redesigned emissions dashboard for AWS customers that allows users to access carbon footprint data directly, without needing IT or finance account permissions. This is a welcome step forward, though the tool’s exclusion of AI-specific metrics is a notable gap at a time when AI energy use is under growing scrutiny.
Investors press Amazon, Microsoft and Google on water and power use at data centres
Shareholders are pushing major tech companies for greater transparency on data centre water and energy consumption ahead of spring annual meetings, with some filing formal resolutions over unmet climate commitments.
Google has made its SpeciesNet AI model freely available to wildlife conservation organisations, allowing teams to automatically identify animal species from camera trap footage at larger scales and speeds than was previously possible.
Chinese data centre operator trials cooking oil-based renewable diesel for backup power
GDS, one of China’s largest data centre firms, has launched a pilot using a renewable diesel made from used cooking oil as a drop-in replacement for conventional diesel in its backup generators for the first time.
The £2 billion Wapseys Wood development in Buckinghamshire has become the first data centre accepted for consideration under the nationally significant infrastructure regime. The proposal includes an on-site gas turbine estimated to emit around 500,000 tonnes of CO2 annually (roughly 0.14% of total UK emissions) putting the government’s AI ambitions on a collision course with its climate commitments and reflecting a wider industry trend of bypassing congested electricity grids in favour of on-site gas.
Deutsche Telekom reports 65% energy saving in 5G network through cloud-based architecture overhaul
Deutsche Telekom says initial live tests of a new cloud-based 5G core network architecture have cut energy use by up to 65%, by dynamically allocating computing resources based on real-time demand rather than running components continuously.
AI to map vulnerable deep-sea ecosystems across the Atlantic for the first time
A UK-led research project will use artificial intelligence to analyse decades of unexamined seafloor imagery and produce the first comprehensive biodiversity maps of vulnerable marine ecosystems across the Atlantic’s High Seas, with findings intended to inform marine protection decisions under the recently adopted High Seas Treaty.
The National Energy System Operator has expanded its Demand Flexibility Scheme to reward households and businesses for increasing electricity use during periods of excess renewable supply, such as sunny weekends, helping to prevent surplus green energy from going to waste.
Academic papers
Now for a summary of some of the most important new digital sustainability research:
Authors: Yuelin Han, Pengfei Li, Adam Wierman, Shaolei Ren
A timely and practically grounded paper that shifts the data centre water debate away from total consumption figures and towards a largely overlooked bottleneck: peak water withdrawal capacity. The core argument is that it is not how much water data centres use on average that matters most, but whether local public water systems can physically supply enough water on the hottest days of the year when cooling demand and water stress peak simultaneously.
Environmental (in)considerations in the design of smartphone settings
Authors: Thomas Thibault, Léa Mosesso, Camille Adam, Aurélien Tabard, Anaëlle Beignon, Nolwenn Maudet Published in LIMITS ’25: Computing within Limits
A rigorous HCI research paper that examines how well smartphone operating systems and applications support users who want to reduce the environmental impact of their digital habits. The paper finds that current settings design actively works against this goal through six recurring anti-patterns, including that environmentally relevant settings are not presented as such, settings provide little information about their actual environmental effects, and default settings are almost always the most resource-intensive option
Authors: Tianqi Xiao, Francesco Fuso Nerini, H. Damon Matthews, Massimo Tavoni, Fengqi You
A rigorous modelling study published in Nature Sustainability that provides one of the most comprehensive assessments to date of the combined energy, water and carbon impacts of AI server deployment across the United States, and reaches sobering conclusions about the industry’s ability to meet its own net-zero commitments.
Assessing the carbon footprint of language models: Towards sustainability in AI
Authors: Fleur Jeanquartier, Claire Jean-Quartier, Paul Rieder, Vedad Misirlić, Christian Pasero, Richard Hohensinner, Heimo Müller, Andreas Holzinger
A benchmarking and transparency-focused paper that evaluates energy consumption during both training and inference for small language models, and finds that the broader AI field has a significant reporting gap and publications simply do not include energy consumption data at all.
Toward a more effective and comprehensive assessment of data center environmental impact
Authors: Dina Nassar, Rabih Bashroush
A methodologically comprehensive paper from the University of East London that develops a new environmental impact model for data centres and applies it to expose significant misalignments in the LEED certification framework, finding that higher certification levels do not reliably correspond to greater emissions savings, and that some of the most impactful sustainability measures are among the least incentivised.
The data heat island effect: quantifying the impact of AI data centers in a warming world
Authors: Andrea Marinoni, Erik Cambria, Luca Dal Zilio, Weisi Lin, Mauro Dalla Mura, Jocelyn Chanussot, Edoardo Ragusa, Chi Yan Tso, Yihao Zhu, Benjamin Horton
A striking paper that introduces a previously unnamed phenomenon, the “data heat island effect,”using satellite-based land surface temperature measurements to quantify the localised warming caused by AI data centres on surrounding communities and environments.
Authors: Olya Kudina, Nynke van Uffelen, Lode Lauwaert, Wim Landuyt Published in AI & Society
A fantastic scoping review from Delft University of Technology and KU Leuven that examines not just what the literature says about AI’s environmental impact, but how it frames the problem, who it assigns responsibility to, what solutions it proposes and, crucially, what ethical assumptions underpin all of this. The findings reveal a field dominated by technical disciplines and a narrow anthropocentric ethics that leaves fundamental questions largely unasked.
Digital sustainability articles
Here are a selection of our favourite articles on digital sustainability from the last month. Click on the title link to be redirected to the full article:
What you need to know before emissions regulators come knocking
Jay Dietrich, writing for Uptime Institute via Computer Weekly, argues that while carbon emissions reporting is rapidly becoming a legal requirement for IT operations globally, the quality and transparency of data provided by cloud and colocation providers remains inconsistent
The life and death of Microsoft’s moonshot
Ketan Joshi delivers a forensic dissection of Microsoft’s climate commitments, arguing that the company’s headline claim of a 23% emissions increase since 2020 obscures a far starker reality that the increase is closer to 55%. Meanwhile, Joshi’s analysis suggests the energy required to generate a million dollars of Microsoft revenue has risen 60% since 2020.
China’s climate focus is shifting from carbon cuts to green tech dominance
Neil Thomas, reporting for The Wire China, argues that China’s climate strategy is undergoing a significant pivot from near-term emissions reduction targets to establishing global dominance in clean technology industries. Thomas argues that this will have major implications for international climate diplomacy, trade relations and decarbonisation timelines.
“AI can help”: American Airlines’ newest solution to get rid of contrails
Daniel S Osipov, in Simple Flying, reports on American Airlines’ collaboration with Google to use AI-generated forecast maps to help pilots avoid creating contrails, which are one of the aviation industry’s most significant contributions to climate change.
Our biggest digital sustainability moment yet
Lydia Tabbron, Digital Sustainability Lead at Defra, reflects on the second Government Digital Sustainability Alliance Summit, which brought together digital and sustainability leaders from government, industry and academia for a day of announcements, panel discussions and the first GDSA Awards ceremony.
What do we mean by digital sustainability?
Catherine Mulligan argues that the dominant understanding of digital sustainability, focused narrowly on energy use, carbon footprints and offsets, addresses only the surface of the problem, and that a genuinely useful framework must balance economic, social and environmental outcomes simultaneously.
AI uses a lot of energy. Some of it is our fault.
Andrew Winston, writing in his Substack newsletter, argues that while AI’s energy footprint is primarily driven by infrastructure and corporate decisions, individual and organisational prompting habits do matter.
Data centres in space: Can they solve AI’s sustainability problem?
Fedor Sukhoi, writing for Impakter, examines whether space-based data centres, which are increasingly promoted by Google, Nvidia and others as a solution to terrestrial energy and land constraints, can genuinely deliver on their sustainability claims, and finds that the published evidence points firmly in the opposite direction once the full lifecycle is accounted for.
Podcast pick
Catalyst with Shayle Kann: The rise of flexible data centres
This month’s pick is from the excellent podcast Catalyst. Host Shayle Kann speaks with Varun Sivaram, CEO of Emerald AI, about the challenge of making data centre infrastructure flexible enough to work with the grid rather than against it. The conversation covers how aligning utilities, cloud providers and grid operators could unlock over 100 gigawatts of unused capacity, and gets into the practical mechanics of how that might actually work, from battery storage to tiered power models.
You can listen to the episode here.
Get involved
We want to hear from you! Share your comments, suggestions, and digital sustainability highlights. Contact our Subject Specialist for Digital Sustainability, Cal Innes, at cal.innes@jisc.ac.uk
And don’t forget to subscribe to our DIGITAL-SUSTAINABILITY JiscMail mailing list for future editions of the Digital Sustainability Newsletter.
