3 min read
The sustainability challenge of AI: Data, Cloud and FinOps strategies
Digital sustainability moves up and down the priority scale for both global organisations and governments alike. But with surging electricity demand and consumption from AI usage and associated data centres now consuming 1.5% of global electricity — electricity consumption of data centres is growing at 12% annually — this issue has never been more urgent.
Which leads us to the challenge that all organisations need to consider as their AI usage matures: how to address environmental impact without compromising business performance and the infinite quest for growth?
The answer is not to stop using technology, it lies in treating sustainability in the same way that financial operations teams measure cloud spend. Getting ahead of the curve of the inevitable regulations that will follow.
The moral obligation matters. The business driver is faster.
What digital sustainability actually means
Digital sustainability is the practice of designing, building and operating technology in a way that reduces environmental harm while still delivering business value.
Covering the physical infrastructure that powers digital services, the way systems are architected, and the technology choices organisations make, we need to face the fact that the acceleration of electricity consumption is driven primarily by AI workloads, which are expected to account for 44% of total data centre power usage by 2030 (Gartner).
The impact is created across three connected areas:
-
Physical infrastructure:
the servers, cooling systems and power supplies operating in data centres.
-
Architectural design:
the way systems are built, how long they run and whether compute resources actually shut down when not needed.
-
Technology selection:
the choice between competing platforms may carry vastly different environmental consequences.
Most organisations do not measure any of these dimensions. Teams optimise for speed and cost without visibility into environmental impact. That gap matters, because the same inefficiencies that waste money also waste energy: idle servers, oversized instances, duplicated data, unnecessary storage and workloads running in regions with higher emissions.
Unsurprisingly, customers are increasingly demanding transparency in this area, and regulators are beginning to force the issue.
Regulatory drivers and increasing pressure
UK public procurement already requires central government buyers to take carbon reduction into account for major contracts, and sustainability expectations are becoming more explicit across digital, cloud and ICT purchasing. The Government Digital Sustainability Alliance, led by Defra, is also shaping guidance and standards for sustainable digital practice across government and its supply chain.
Private-sector organisations face similar pressure from investors scrutinising environmental, social and governance (ESG) commitments, and from end consumers who increasingly expect corporate accountability on sustainability.
This vast energy consumption by companies competing in the AI space is not a fringe concern discussed only at sustainability conferences — it surfaces repeatedly in conversations with customers who care about their environmental footprint, whether from genuine conviction or regulatory necessity.
Measurement, FinOps and better decisions
The effective business argument: cost optimisation and environmental optimisation are largely the same problem.
FinOps for cloud focuses on controlling cloud spend through intelligent resource management. Visibility into consumption often reveals inefficiencies that cost energy as well as money: servers running 24/7 when they could be shut down overnight, redundant resources kept live unnecessarily, or oversized instances purchased to handle peak load despite running well below capacity most of the time.
The overlap between cost and environmental optimisation is substantial. Shutting down compute that is not needed saves both money and the planet. Choosing a data centre region with renewable energy reduces emissions at no additional cost.
Action steps for your organisation
-
1. Measure first
Audit your current cloud consumption. Establish a baseline and track progress. Without measurement, you cannot improve or communicate progress to stakeholders.
-
2. Build sustainability into architecture
Design reviews should consider environmental impact alongside performance, resilience, security and cost.
-
3. Use FinOps as the operating model
Bring carbon and water metrics into cloud optimisation conversations, so cost savings and sustainability improvements are pursued together.
-
4. Make it visible
Put environmental data where teams see it regularly, using metrics that are meaningful to both technical and business stakeholders.
The bottom line
Digital sustainability is not about environmental purity. It is about balancing the business need for effective, scalable infrastructure with the legitimate environmental cost of that infrastructure. The goal is to reduce that cost through smarter design, better measurement and deliberate choices – not to eliminate technology, but to use it more thoughtfully.
Organisations that move first on this will find themselves ahead when regulation tightens and customers begin expecting proof of environmental stewardship. Those that wait will face scrambling retrofits and reputational risk.
How we can help
Digital sustainability requires measurement, strategy and deliberate choices about how you build and operate infrastructure. We work across three core areas to embed environmental responsibility into technical delivery:
-
Green Ops
is our environmental sustainability offering within our managed service solutions. The EcoDelve dashboard translates cloud consumption into relatable metrics: the number of showers powered by data centre cooling, the trees needed to offset carbon, the distance you could drive. As we discussed earlier, rather than abstract kilowatt hours, this grounding makes sustainability visible to both technical teams and stakeholders. We measure Emissions Level Agreements (ELAs) alongside traditional Service Level Agreements, ensuring environmental performance is built into service delivery from the start – not retrofitted later.
-
FinOps
is the practice we have deployed extensively to help teams control cloud spend. FinOps for cost optimisation identifies the same inefficiencies that waste energy: unnecessary instances, resource bloat, suboptimal region choices: the same work that cuts costs cuts environmental impact. FinOps for AI addresses the specific challenge of sustainable AI workload deployment as AI models consume extraordinary power during training and inference.
-
Architecture and design
is where environmental responsibility becomes structural. Our teams integrate sustainability into design reviews, making environmental footprint a first-class constraint alongside performance and cost.
Ready to act?
If your organisation needs help auditing consumption, designing sustainable systems or implementing emissions tracking, talk to us to discuss where digital sustainability fits in your roadmap.
You can also learn more about our Unified Data and AI offering.
About the author
Dawn is a strategic data, analytics and AI leader, with over 20 years’ experience helping organisations turn data into business value. Her expertise spans data strategy, governance, analytics, data and AI adoption and enterprise platform delivery, with a strong track record of leading large-scale transformation programmes across global organisations. Combining deep technical knowledge of Microsoft Azure, Microsoft Fabric, Databricks, AWS, Snowflake and Oracle with extensive leadership experience, Dawn is passionate about building high-performing teams and enabling organisations to realise the full potential of their data assets.
