If you haven’t already, be sure to subscribe to our DIGITAL-SUSTAINABILITY JiscMail mailing list to receive future editions of the Jisc Digital Sustainability Newsletter directly to your inbox.
Editor’s Note
Welcome to the latest edition of the Digital Sustainability Newsletter.
You might notice that this month’s issue has a strong focus on data centre energy use, reflecting how rapidly the debate over powering our digital infrastructure has intensified. With AI driving exponential growth in computation, questions around efficiency, sufficiency, and accountability have become recurring themes across digital sustainability news and research.
You’ll also see a new addition this month: the ‘Academic Papers’ section. While we’ve always included key research in previous editions, it is felt that the pace and importance of recent scholarly work in this area now warrant a dedicated space.
The aim of this newsletter remains the same: to bridge the gap between academic insight and real-world application. Whether you’re an IT leader in further education, a post-doctoral researcher exploring sustainable computing, or simply someone curious about how technology and climate goals align, the Jisc Digital Sustainability Newsletter is designed to bring those worlds together.
I hope you enjoy this slightly new direction. As ever, if you’re working on research or initiatives that could feature in a future issue, please do get in touch.
— Cal Innes, Digital Sustainability Specialist, Jisc (cal.innes@jisc.ac.uk)
Digital Sustainability News
From government initiatives to emerging technologies, here are some of the latest updates driving progress in digital sustainability. Click the headline to read more.
King’s College London develops tool to measure carbon footprint of computational research
Researchers at King’s College London have created a tool to estimate the greenhouse gas emissions of computational research using high-performance computing facilities. The project aims to help scientists understand both energy and manufacturing-related emissions from computing, encouraging more sustainable research practices across disciplines.
Defra team develops digital sustainability assessment
Defra’s digital sustainability team has created a new assessment process for evaluating the environmental performance of government digital services. The framework, designed to support both service teams and assessors, introduces a three-part standard: a sustainability risk evaluation early in development, a sustainability statement updated before service assessments, and a guide with criteria for red–amber–green ratings. The department plans to integrate the system into service assessments in early 2026 and share it with other government departments.
Microsoft to offer free Windows 10 security updates in Europe after consumer pressure
Microsoft will provide free extended security updates for Windows 10 users in the European Economic Area without requiring them to enable Windows Backup, following pressure from consumer group Euroconsumers. The move comes ahead of Windows 10’s end of support in mid-October and applies only within Europe, with users elsewhere still needing to pay or use Microsoft’s backup service.
Study finds emissions from 10 new UK data centres could cancel 2025 gains from electric vehicles
Research by Foxglove and Global Action Plan suggests that just ten of the large data centres planned for the UK could emit around 2.7 million tonnes of CO₂e annually, effectively offsetting the year’s carbon savings from the public’s switch to electric cars. The findings highlight uncertainty and poor transparency around the true climate impact of Britain’s expanding digital infrastructure.
The Department for Environment, Food and Rural Affairs has published a five-year digital sustainability strategy aiming to cut technology-related carbon emissions by 16% and ensure 80% of staff devices are refurbished by 2030. The plan also requires major suppliers to commit to verified net zero targets and introduces stricter sustainability reporting and social value criteria for contracts.
Google’s huge new Essex datacentre to emit almost 570,000 tonnes of CO2 a year
Planning documents for Google’s 52-hectare “hyperscale” datacentre in Thurrock, Essex, estimate operational emissions of 568,727 tCO₂e per year, as the UK accelerates AI computing capacity. Ministers argue grid decarbonisation will limit national-budget impacts, but campaigners warn of added pressure on emissions, power and water.
Europe approves first onshore carbon storage permit using Icelandic mineralisation technology
The EU has granted its first permit for onshore geological CO₂ storage to Icelandic firm Carbfix, marking a major step toward land-based carbon sequestration. The company’s technology turns captured CO₂ into solid rock within two years, offering a low-leakage, nature-inspired method that could help hard-to-abate industries cut emissions. The firm is projected to store 100,000 tonnes of CO₂ annually, with a goal of 3.2 million tonnes over 30 years.
Google removes public online reference to 2030 net-zero goal amid rising AI energy use
Google has quietly deleted mention of its 2030 net-zero emissions target from its main sustainability webpage, despite insisting the pledge still stands internally. The change comes as the company’s emissions have risen nearly 50% in a year due to surging AI data-centre power demand and as political pressure on climate action grows in the US.
Data centre boom in drought-hit Mexican state raises water and energy concerns
Querétaro, Mexico’s emerging data centre hub attracting billions in investment from firms such as Microsoft and Amazon, is facing criticism for the industry’s heavy water and energy use amid severe drought conditions. Local activists say industry expansion is worsening water scarcity, while operators insist they comply with environmental rules.
Startup’s technology eliminates up to 90% of methane emissions from dairy barns
Danish startup Ambient Carbon has developed a system called MEPS (short for Methane Eradication Photochemical System) that, when trialled in a semi-enclosed dairy barn, successfully destroyed up to 90% of dilute methane ventilation air, potentially offering a scalable method to cut major dairy farm emissions.
Academic Papers
This brand new section spotlights and summarises recent academic work offering evidence-based insights into the technologies and policies driving a more sustainable digital future.
More than carbon: cradle-to-grave environmental impacts of GenAI training on the Nvidia A100 GPU
Authors: Sophia Falk, David Ekchajzer, Thibault Pirson, Etienne Lees-Perasso, Augustin Wattiez, Lisa Biber-Freudenberger, Sasha Luccioni, Aimee van Wynsberghe
Summary: This groundbreaking study provides the first full life-cycle assessment of AI training on Nvidia A100 chips, tracing environmental impacts from production to operation. It finds that the use phase – when models are actually being trained – accounts for around 96% of total climate impact, while manufacturing contributes more to toxicity and resource depletion. The authors argue that AI sustainability can’t be measured by carbon emissions alone and that wider environmental impacts need much greater attention.
Ground-truthing AI energy consumption: validating CodeCarbon against external measurements
Author: Raphael Fischer
Summary: This new study tests the accuracy of popular AI carbon tracking tools like CodeCarbon, comparing their estimates with real-world measurements from hundreds of machine learning experiments. The research finds that while these tools broadly reflect energy use patterns, they can misestimate AI model energy consumption by up to 40%, underscoring the need for better validation and transparency in sustainable AI reporting.
Authors: Julien Delavande, Regis Pierrard, and Sasha Luccioni
Summary: This study takes a close look at how much energy text-to-video models really use and the results are striking. The researchers find that energy demand increases rapidly as video quality and length go up – even small improvements in resolution can cause big jumps in power use. By benchmarking six leading open-source models, the study highlights the urgent need to make generative video systems far more energy-efficient as the technology scales.
From component to system: rethinking edge computing design through a carbon-aware lens
Authors: Xuesi Chen, Ariel Goldner, Eren Yildiz, Ilan Mandel, Tingyu Cheng, Josiah Hester, and Udit Gupta
Summary: A new framework for carbon-aware edge computing design finds that sustainability depends heavily on context, with the lowest-emission setup varying by workload, power source, and deployment length. The study shows that solar- or battery-powered devices aren’t always the greenest options, as embodied emissions from components like batteries and solar infrastructure can outweigh operational savings, revealing the need for system-level carbon optimisation rather than relying on energy efficiency alone.
Authors: Kejuan Sun, Youfu Yue, Lafang Wang, and Jiabai Ye
Summary: The researchers behind this paper have developed a comprehensive global model that tracks how different forms of digital integration affect carbon emissions across economies and industries. Their analysis of 19 countries shows that digital activities are generally less carbon intensive than traditional sectors, but that downstream integrations (where digital products depend heavily on non-digital inputs) carry higher carbon costs. Meanwhile, midstream digital integration, where digital tools are embedded within production systems, appears to cut emissions significantly in high-impact sectors such as manufacturing and transport.
Imposing AI: deceptive design patterns against sustainability
Authors: Émile Beigon, Thomas Gibon, and Shazade Jameson
Summary: This paper argues that AI’s environmental footprint has reached a scale demanding formal limits, much like those in other high-impact industries. The authors propose a governance framework combining technical thresholds, regulatory caps, and ethical constraints to align AI development with planetary boundaries. Their work reframes the debate from reducing emissions to actively enforcing ecological limits on compute, data, and model deployment.
Not all water consumption is equal: a water stress weighted metric for sustainable computing
Authors: Yanran Wu, Inez Hua, and Yi Ding
Summary: This paper introduces SCARF, a new framework for assessing computing’s water footprint that adjusts for regional and seasonal water stress. Rather than treating all water use equally, the study shows how datacentre cooling, chip manufacturing, and AI workloads can have drastically different environmental impacts depending on where and when they consume water, revealing opportunities to lower impact simply by shifting operations in time or location.
Aligning digital futures with ecological citizenship for sustainability
Authors: Luke Gooding, Robert Phillips
Summary: The authors of this paper propose Ecological Citizenship as a practical compass for a Sustainable Digital Society, shifting digital responsibility from individual “good behaviour” to shared civic duties across governments, industry, and citizens. Through three case studies (Green Digital Charter, SolarKiosk, Fab City), the paper shows how policy, infrastructure, and local production can embed environmental justice and circularity into digital systems.
Energy efficient or exhaustive? Benchmarking power consumption of LLM inference engines
Authors: Chenxu Niu, Wei Zhang, Yongjian Zhao & Yong Chen
Summary: A new benchmarking study dissects the energy draw of popular LLM inference engines (including vLLM, TensorRT-LLM, and DeepSpeed) on dual Nvidia H100 GPUs, showing major variation in both total power and subsystem use. By splitting the inference lifecycle into setup and token-generation stages, the authors reveal that GPU power dominates, while CPU and DRAM still contribute measurable overheads. The paper finds that differences in software design and memory management can swing energy efficiency by more than 30%, highlighting a need for standardised energy reporting in AI deployment.
Optimising microgrid composition for sustainable data centres
Authors: Julius Irion, Philipp Wiesner, Jonathan Bader, Odej Kao
Summary: As data centre energy demand soars, this paper introduces a new optimisation framework for designing on-site renewable microgrids that co-locate solar, wind, and battery systems with data centres. By integrating the Vessim co-simulator with the NREL System Advisor Model (SAM) and coupling it to the Optuna optimisation engine, the authors simulate how different microgrid compositions affect operational vs. embodied carbon emissions. Applied to real workload traces from the Perlmutter supercomputer, the tool identifies region-specific trade-offs: Berkeley achieves near-zero operational emissions at lower embodied cost, while Houston requires larger investments to reach similar decarbonisation levels. The study provides practical guidance for carbon-aware infrastructure planning and highlights that “net-zero” configurations aren’t always optimal once lifetime embodied emissions are considered.
Resource corner
Each month, we share a digital sustainability report, tool, or resource that we think is worth your time.
How data centre expansion risks derailing climate goals
This month’s featured resource is a new report from ECOS, developed in collaboration with Open Future, examining how the rapid growth of data centres could undermine Europe’s climate goals.
The report, How Data Centre Expansion Risks Derailing Climate Goals (and How to Fix It), looks at the full environmental footprint of data infrastructure: energy use, emissions, material extraction, and water consumption. It argues that while efficiency improvements are important, they are not enough. Instead, the authors call for a broader shift in how digital infrastructure is planned and governed, embedding sufficiency, circularity, and transparency into its design and regulation.
Key points include:
- Data centre emissions may be more than six times higher than publicly disclosed.
- The rise of AI is intensifying pressure on energy systems and material supply chains.
- Existing EU measures focus too narrowly on efficiency, overlooking sufficiency and accountability.
- A new policy framework is needed to balance innovation with environmental limits.
The report makes a strong case for rethinking the digital transition—not as an endless expansion of computational capacity, but as a system that must stay within planetary boundaries.
You can download the full report and infographic, and watch the accompanying webinar, via ECOS.
Digital sustainability articles
Here is 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 kind of environmental impacts are AI companies disclosing? (And can we compare them?)
Researchers Sasha Luccioni and Theo Alves da Costa analyse in their Hugging Face blog how AI companies such as OpenAI, Mistral and Google are publishing figures on their models’ energy, water, and carbon use, but argue that inconsistent methodologies and missing totals make the disclosures impossible to compare.
Writing on the UK Government Sustainable ICT blog, Rich Kenny, Avinash Lunj and Alexandra Kis from the Government Digital Sustainability Alliance warn that artificial intelligence could drive global water use from 1.1 billion to 6.6 billion cubic metres by 2027 – more than half of the UK’s annual consumption. They explain that while GPT-3 alone used around 700,000 litres during training, most AI cooling water can’t be reused, raising serious concerns for biodiversity and water-stressed regions.
In Harvard Magazine, Olivia Farrar reports on a Harvard Climate Action Week panel that examined the true environmental costs of artificial intelligence, from the 4% of U.S. electricity already used by data centres (mostly from fossil fuels) to the public health and community impacts of siting power-hungry AI infrastructure. The panel’s experts called for a shift toward “Green AI” practices that prioritise energy efficiency, emissions transparency, and equitable development over scale and speed.
How do you build a more sustainable data centre?
Bianca Giacobone’s report for Latitude Media, offers new insights from Ed Ansett of Ramboll, whose white paper “Developing sustainable data centres” outlines practical strategies to cut the industry’s environmental impact. While data centres remain energy- and carbon-intensive, Ansett argues that measures such as renewable energy sourcing, heat reuse, and circular design could significantly reduce their footprint, though cost pressures still challenge widespread adoption.
IT sustainability think tank: how IT directors can spot false green claims from big tech suppliers
In an article from Computer Weekly, Craig Wentworth of TechMarketView argues that as Big Tech ramps up sustainability marketing, IT directors must develop the skills to tell genuine environmental performance from greenwashing. He warns that many suppliers use vague averages, shifting pledges, and excessive reliance on carbon offsets to obscure the true footprint of their data centres and AI products, and says that verifiable, present-tense emissions data should be the benchmark for trust.
Everyone’s talking about data centre flex. Nobody’s talking about clubbing
James Johnston, CEO and co-founder of the flexibility marketplace company, argues that while the industry is rightly excited about “data centre flex” – the ability of large data facilities to shift or curtail their own energy use – few are discussing a complementary model: data centres procuring flexibility from local Virtual Power Plants, a concept he calls “clubbing.” In this approach, data centres share energy resources and capacity with nearby producers, helping to relieve grid congestion, cut costs, and strengthen community resilience through a more collaborative energy ecosystem.
Worried about AI’s soaring energy needs? Avoiding chatbots won’t help – but 3 things could
In ZDNET, Radhika Rajkumar and Sabrina Ortiz explore whether AI use is as environmentally harmful as headlines suggest. They find that while data centres now consume about 1.5% of global electricity and could reach 7.5% of US power use by 2030, the impact of an individual ChatGPT query is relatively minor. Instead of boycotting AI, they argue that users and companies should focus on transparency, sustainable procurement, and using smaller models to meaningfully reduce the technology’s footprint.
AI: Five charts that put data-centre energy use – and emissions – into context
In Carbon Brief, Josh Gabbatiss uses five data-rich charts to illustrate how artificial intelligence is reshaping global energy demand through the rapid growth of data centres. Drawing on new International Energy Agency figures, he shows that while data centres currently account for just over 1% of global electricity use and 0.5% of CO₂ emissions, their expansion could drive up to 12% of the world’s electricity demand growth by 2030. Gabbatiss explains that although clean energy is scaling faster than fossil fuels, data-centre emissions remain one of the few areas still rising, posing fresh challenges for global net-zero targets.
In a Q&A with ESG Dive, Lauren Schenkman interviews Jack Roswell, co-founder and CEO of Perennial, about how the company’s new VT0014 soil carbon quantification software could make carbon offsetting more credible and affordable. Roswell explains that the software, which is now certified by Verra, one of the voluntary carbon market’s most rigorous standards, uses AI modelling, environmental data, and satellite imagery to estimate how much carbon soil captures, requiring far fewer soil samples. Through a mix of visuals and clear explanations, the piece highlights how digital mapping and modelling tools can help restore trust in a market long criticised for weak verification.
How AI can transform sustainability reporting
Sacha Bazin and Mike Hayes in the World Economic Forum, explain how artificial intelligence is reshaping corporate sustainability reporting, helping organisations automate data collection, validate disclosures, and improve visibility across complex supply chains. The authors note that while AI tools can increase accuracy and reduce administrative costs, their effectiveness depends on transparency, model validation, and continued human oversight. Used responsibly, AI could make sustainability disclosures faster and more reliable, but poorly implemented systems risk amplifying errors and undermining trust.
Green energy microgrids hailed as cost-effective answer to UK’s datacentre energy supply woes
Writing for Computer Weekly, Caroline Donnelly reports that new research from the Centre for Net Zero suggests renewable microgrids could offer a cheaper, faster, and more scalable alternative to nuclear small modular reactors (SMRs) for powering the UK’s growing network of data centres. With AI and cloud workloads driving record demand for electricity, the study argues that microgrids using on-site solar, wind, and battery storage can be deployed more rapidly than nuclear projects, at significantly lower cost per megawatt-hour, potentially cutting both emissions and energy bills for operators.
Podcast pick
Each month we highlight a podcast episode that brings fresh insight into digital sustainability and climate tech.
The climate question: will the switch to green technology become unstoppable?
This month’s podcast recommendation comes from The Climate Question by the BBC World Service. Our chosen episode asks a hopeful yet urgent question: Will the switch to green technology become unstoppable?
Hosts Graihagh Jackson and Jordan Dunbar speak with Professor Tim Lenton, Director of the Global Systems Institute at the University of Exeter and author of Positive Tipping Points: How to Fix the Climate Crisis. Lenton argues that while we often hear about negative tipping points, like melting ice sheets or collapsing ecosystems, there are also positive tipping points where climate solutions can suddenly accelerate and become self-sustaining.
He discusses real-world examples of these turning points already taking place, from the rapid global uptake of renewables to the explosion in electric vehicle adoption, and even the surprising role that 1980s pop band A-Ha played in kickstarting Norway’s EV revolution.
It’s a fascinating, optimistic look at how social, technological, and political feedback loops might be shifting in our favour, and what that means for the speed of the green transition.
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.
