In summary:
- AI’s hidden environmental impact is significant. From rare earth mining to energy-hungry data centers, unchecked AI growth threatens ecosystems, water resources and global net-zero goals.
- Sustainable AI solutions are essential for progress. Andrew Grigg's STAR framework helps organisations measure, reduce and manage AI’s footprint while using AI to drive climate action and improve resource efficiency.
- Collaboration is the key to building a greener future. Government and industry initiatives like the GDSA enable organisations to benchmark AI tools, share best practices and embed sustainability into digital transformation.
Artificial Intelligence (AI) is reshaping how we work, live and govern. But beneath the surface of this technological revolution lies an often-invisible environmental footprint, one that is frequently underestimated, underreported, and rapidly growing.
We believe digital innovation must go hand-in-hand with environmental responsibility. That’s why we’re working with government partners to uncover the true impact of AI and to develop practical frameworks that reduce its cost to the planet.
The environmental cost of intelligence
AI’s environmental impact spans the entire value chain, from mining rare earth materials to energy-intensive training of large language models (LLMs), through to the huge volumes of water consumed by data centres. According to the International Energy Agency (IEA), electricity demand from data centres could double by 2026, driven largely by AI workloads. Many of these facilities are built in water-stressed regions, putting added pressure on local resources.
If left unchecked, this growth could have catastrophic consequences for the environment. Rising emissions from fossil-fuelled energy grids risk undermining global net-zero targets, while the millions of litres of freshwater consumed to cool AI systems threaten to intensify droughts and reduce agricultural productivity. The scale of extraction for rare earth minerals required to power AI infrastructure also poses a significant risk to fragile ecosystems, driving biodiversity loss and generating record levels of toxic e-waste, as highlighted in the United Nations Global E-waste Monitor 2024.
Put simply, AI development could transform a technology of promise into a driver of environmental collapse. That’s why it is vital that governments and industry prioritise sustainability to ensure AI contributes to a greener, more resilient future.
So, before you consider using AI you should:
understand and manage the environmental impact of the AI systems you are planning to use.
use AI technologies only when relevant, appropriate and proportionate.
choose the most suitable and sustainable option for your organisation’s needs.
A dual mission to sustainable AI
Sustainable AI is not just about reducing the environmental footprint of AI systems, it’s also about using AI to advance sustainability. As Gartner explains, “the trade-offs lie in combining ‘AI for sustainability’ with the ‘sustainability of AI’.”
Both are essential. AI can accelerate climate action and improve resource efficiency, but without strong measurement and accountability, efforts to reduce the footprint of AI itself risk being overlooked. To achieve real progress, organisations must balance innovation with responsibility, ensuring that every AI initiative considers both the benefits it delivers and the resources it consumes.
Turning insight into action with the STAR framework
To help organisations make tangible progress, we have developed the STAR framework, a four-stage approach to sustainable AI:
Strategise: define where AI supports sustainability goals and where its footprint must be reduced.
Tally: measure impacts using tools such as Hugging Face’s AI Energy Score, EcoLogic, and Sopra Steria’s own EcoMind AI.
Act: apply best practice across the AI lifecycle, from model selection to data centre location. For example, shifting workloads from German to French energy sources (more nuclear-based) reduced emissions sevenfold in one of our projects.
Review: continuously monitor, report, and refine.
This framework is already informing government practice, such as The AI Playbook for the UK Government.
Collaboration is key
No single organisation can address the environmental impact of AI alone. The UK Government is already taking steps through initiatives such as the Government Digital Sustainability Alliance (GDSA), which brings together industry, government and academia to accelerate progress. Within this, a dedicated working group on AI sustainability is helping to shape reporting standards, share best practice, and influence supply chains towards greener solutions.
Collaboration is about practical action. By partnering across sectors, we can benchmark the environmental impact of AI tools and ensure that sustainable practices are built into the foundations of digital transformation. But more voices are needed, so what can you do today?
Raise awareness by starting conversations with colleagues, stakeholders and suppliers about AI’s hidden footprint.
Apply the STAR framework: begin with measurement, even if the data is imperfect.
Share resources such as The AI Playbookfor the UK Government.
Collaborate: Sopra Steria is inviting partners to benchmark open-source LLMs using EcoMind AI.
Looking ahead
AI is not a passing trend; it is a defining force of our time. But unchecked growth comes at a high environmental cost.
We may not yet have all the answers, but we have the frameworks, the partnerships and the determination to act. As Andrew Grigg recently noted: “Let’s not let perfection get in the way of progress”. The future of AI must be sustainable and that future starts now.
Get in touch with Andrew Grigg and partner with us to benchmark AI tools, co-create greener solutions, and ensure sustainability is built into every stage of AI development.