Sustainable cloud computing, often called green cloud computing, means running cloud workloads with less energy, carbon, water and hardware for the same work, and tracking that impact at the same level of detail as the cloud bill. It is the cloud part of sustainable IT.
Large cloud data centres spend less energy on cooling and power overheads than the industry average, but digital growth means demand for that digital infrastructure is rising fast. The International Energy Agency (IEA) reports that data centre electricity demand rose 17% in 2025 and expects it to double by 2030. Moving workloads to the cloud does not cut emissions by itself. The reductions come from how teams size, place, schedule and measure those workloads, and most of the same changes lower the cloud bill.
This article covers what makes cloud computing more or less sustainable, the changes that make up a sustainable cloud strategy, and how to measure cloud sustainability over time. Our guide to what GreenOps is covers the wider practice of managing cloud cost and carbon together.
What is sustainable cloud computing?
Green cloud computing and sustainable cloud computing are two names for one goal, cutting the environmental impacts of cloud services over their whole life. Those impacts have four parts:
- electricity used by servers, storage and network equipment, plus the data centre's cooling and power overheads
- greenhouse gases emitted when that electricity is generated, which depends on the local grid
- water used to cool data centres and to generate their electricity
- embodied carbon from manufacturing the hardware before it is switched on
Providers control the physical infrastructure, including buildings, cooling and energy contracts, while customers control what they run, how big it is, where it runs and when. A sustainable approach to the cloud works on the customer side, where teams can make changes and see the effect on their own bill.
Some guides use "green computing" or green information technology for an organisation's laptops, printers, office hardware and on-premises information systems. This article covers public cloud services from Amazon, Microsoft and Google, the tech companies behind AWS, Microsoft Azure and Google Cloud. The same principles apply to edge computing.
Is cloud computing sustainable?
Cloud computing can be more efficient than running the same workloads on-premises, mainly because hyperscale data centres spend less energy on cooling and power conversion. Power usage effectiveness (PUE) measures that overhead as total facility energy divided by IT equipment energy. The Uptime Institute's 2026 survey put the industry average at 1.52, against published fleet averages of 1.14 for AWS in 2025, 1.17 for Microsoft Azure in its financial year to June 2025, and 1.09 for Google.
Data centre PUE: industry average vs the big three clouds
Bars show the overhead: extra energy for cooling and power for every 1 kWh used by IT equipment. Lower is better.
| Data centres | PUE |
|---|---|
| Industry average (Uptime Institute, 2026) | 1.52 |
| Microsoft Azure (Financial year to June 2025) | 1.17 |
| AWS (2025) | 1.14 |
| Google (2025) | 1.09 |
Cloud growth and the future of data centre infrastructure
A 2020 study in Science found that efficiency gains kept global data centre energy use roughly flat through the 2010s while demand for computing grew quickly with the spread of digital technologies. Since then, demand has grown faster than energy efficiency has improved, and forecasts for the near future point the same way. The IEA expects electricity use by AI-focused data centres to triple by 2030. Google's 2026 environmental report records a 37% annual increase in its electricity demand, and Amazon's 2025 sustainability report says its absolute carbon emissions rose in 2025.
A rightsized service in an efficient region with a clean grid is a lower-carbon choice than an idle server in a company basement, while an oversized estate left running around the clock on a grid that relies on gas and coal can be worse than either. Our article on data centre energy consumption has the full demand forecasts.
Renewable energy claims from cloud providers
Amazon says in its 2025 sustainability report that it has matched 100% of the electricity consumed by its global operations with renewable energy for three consecutive years. Google's 2026 environmental report says it has matched 100% of its electricity consumption with renewable energy purchases for the ninth year in a row.
Matching is an annual accounting measure, in which a provider buys renewable electricity or certificates equal to its total use over the year, while each data centre still draws from its local grid hour by hour, including at night and when the wind drops. Google tracks that gap separately through its 24/7 carbon-free energy goal, which it calls a moonshot in the same report.
For cloud customers, the difference shows up in reporting. The GHG Protocol Scope 2 Guidance sets out two ways to report emissions from purchased electricity. A market-based figure reflects the provider's contracts and renewable energy certificates. A location-based figure reflects the grid where the server runs. The two can differ by a wide margin, and our entry on location-based vs market-based emissions explains when each applies. A move to a cleaner region shows most clearly in the location-based figure.
Sustainable cloud in practice
Most reductions come from sustainable practices that teams adopt inside their own accounts, and many of those changes also cut cost. FinOps and cloud operations teams often already run them for cost reasons.
Remove idle and oversized resources
Every idle virtual machine, forgotten web server, unattached disk and oversized database draws power and produces nothing. Removing idle resources and rightsizing the rest is where most teams start, because the saving shows on the bill and in the footprint at the same time. AWS Compute Optimizer and Azure Advisor flag candidates, and non-production environments can be scheduled to shut down outside working hours. Autoscaling, a form of dynamic resource allocation, adds and removes capacity as load changes. Our cloud cost optimisation guide covers the workflow step by step.
Choose cleaner regions
The same instance type has a different footprint in different regions, because each region draws on a different grid. Where latency, data residency and price allow, placing new or flexible workloads in regions with cleaner grids cuts location-based emissions with no code changes.
Price, carbon and water can each point to a different region. In Greenpixie's Region Scorecard, AWS eu-north-1 in Sweden scores A+ for carbon and is in the cheapest price band. AWS eu-central-2 in Switzerland also scores A+, but is in the most expensive band, and eu-south-2 in Spain is mid-priced and scores A for carbon but faces extreme water stress.
In Greenpixie's estimates for November 2026, the same AWS m5.12xlarge instance running for a year produces about 620 kg CO2e in Hong Kong's ap-east-1 region and 284 kg in Spain's eu-south-2, 54% less. Its electricity use differs far less, 1,258 kWh against 1,067 kWh, so most of the gap comes from the local grid.
Carbon aware computing
Carbon aware computing means running flexible work at the times and in the places where electricity is cleanest. The measure behind it is carbon intensity, the greenhouse gases emitted per kWh from a grid, which changes by region and by hour as the mix of wind, solar, gas and coal shifts. Batch jobs, model training, backups and CI pipelines are common candidates, because DevOps engineers can let them wait a few hours or run elsewhere.
In 2020 Google began shifting the timing of compute tasks within its data centres to the hours when wind and solar power were most plentiful, as it described at the time. The Green Software Foundation's open-source Carbon Aware SDK lets teams build the same behaviour into their own applications and deployment pipelines. Greenpixie's GPX Grid supplies hourly carbon intensity for 269 cloud regions across AWS, Azure and Google Cloud, built from 91 grid areas, so a team can estimate the saving from a time or region shift for each workload.
Choose energy-efficient technology and services
Hardware innovation means newer instance generations and other processor families can complete the same work with less energy, and shared managed services can make more efficient use of hardware than a dedicated server. The size of the gain depends on the workload, so the practical test is to compare options on energy and CO2e per unit of usage, in the same way teams compare price per hour. A carbon rate card lists the footprint of each service and SKU, so cloud architects and engineers can weigh it alongside cost when they choose technology and design the architecture for a migration or digital transformation project.
Store and move less data
Stored data uses power and hardware for every hour it is kept, and every GB moved between regions uses network energy. Deleting unused snapshots, moving cold data to archive tiers and keeping chatty services in the same region cut both. Greenpixie estimates data transfer within a region and between regions separately, per GB, so these savings appear in the same report as compute savings.
AI inference is a fast-growing share of cloud use with its own levers, such as model choice, caching and batching. Our sustainable AI guide covers them.
Measuring cloud sustainability and carbon footprint
Teams need a baseline of their cloud environmental footprint before they can track any of these changes. AWS, Azure and Google Cloud each offer a footprint tool, but the figures arrive as monthly totals by service and region, and each provider uses its own method. That makes them hard to compare across providers and too coarse to tie to a single resource or team. Our comparison of cloud carbon tools covers the differences.
Greenpixie estimates energy, CO2e and water for every line of a cloud provider's usage report, such as the AWS Cost and Usage Report. It maps each SKU to the hardware that delivers it, estimates the energy per unit of usage, then derives CO2e and water from the region's grid and data centre factors. The data covers 99%+ of typical customer cloud usage across AWS, Azure and Google Cloud, and lands in the FinOps tools or data warehouse a team already uses. Greenpixie's methodology has been independently verified in accordance with ISO 14064-3, prepared with reference to ISO 14064-1, and designed to be in line with the GHG Protocol Corporate Standard. See GPX Cloud for how it works.
Our guide to GreenOps metrics and KPIs covers which measures to track over time, such as CO2e per unit of business output. For annual reporting under CSRD, SECR or CDP, read our article on Scope 3 cloud emissions.
Cost and carbon savings together
Most sustainable cloud work cuts cost as well, because it reduces the hardware a workload uses and the energy costs built into the price of that hardware. Energy follows cost more closely than CO2e does, since CO2e also depends on the grid. Region moves are the main exception, where the cheapest region and the cleanest one can differ. Our article on FinOps and sustainability covers how to bring both into one practice.
At Mastercard, cloud cost optimisations had stalled because cost savings alone did not create enough urgency, and adding carbon data next to cost gave teams a second reason to act. As our GreenOps page sets out, Greenpixie data enabled $2M in cloud cost reductions, and Mastercard exceeded its cloud carbon reduction target.
James Hall made the same point in a 2025 techUK guest post, writing that "Most people can intuitively figure out that processes like right-sizing & autoscaling are going to save carbon as well as cost, but they don’t know by how much."
Measure your cloud emissions
See the energy, carbon and water behind your cloud bill
GPX Cloud adds energy, carbon and water data to AWS, Azure and Google Cloud billing data, line by line.
Frequently asked questions
What is green cloud computing?
Green cloud computing applies environmental sustainability to cloud services, running them with a smaller environmental impact for the same work. Providers work on efficient data centres and clean energy contracts. Customers work on rightsizing, region choice, scheduling and measurement.
Is cloud storage environmentally friendly?
Cloud storage in a large data centre has lower cooling and power overheads than the industry average, so the same data can have a smaller carbon footprint there than in a small server room. It still uses power, cooling and hardware for every hour it is kept. Deleting unused data and old snapshots, and moving rarely used data to archive tiers, reduce that footprint.
Is AWS 100% renewable?
Amazon reports that it matched 100% of the electricity used by its global operations, which include AWS, with renewable energy in each of the last three years. That is annual matching through contracts and certificates. AWS data centres still draw from local grids hour by hour, so location-based emissions from AWS usage are above zero.
Why is cloud computing also a green solution?
Server virtualization lets cloud providers share hardware across many customers, which raises utilisation, and run data centres with lower cooling and power overheads than the industry average. Moving workloads from underused on-premises servers to rightsized cloud instances can cut their energy consumption. Resources left idle or oversized lose that gain, however much green technology the data centre uses.
Is cloud computing bad for the environment?
Cloud computing uses a fast-growing amount of electricity, and generating that electricity contributes to climate change wherever grids still burn fossil fuels. The IEA reports that data centre electricity demand rose 17% in 2025. Its environmental cost depends on how it is used. Rightsized workloads in efficient data centres on clean grids have a lower footprint than idle or oversized resources on grids that rely on gas and coal.
Is the cloud better for the environment than on-premise servers?
It can be. Hyperscale data centres publish PUE figures between 1.09 and 1.17, against an industry average of 1.52 in the Uptime Institute's 2026 survey, and they share hardware across many customers. The gain depends on rightsizing workloads after a migration and on the grid in the chosen region, so it is worth estimating the carbon footprint before and after a move.
How do I choose a low-carbon cloud region?
Compare regions on grid carbon intensity alongside latency, data residency, price and water stress. In Greenpixie's November 2026 estimates, the same AWS instance produces 54% less CO2e in Spain's eu-south-2 than in Hong Kong's ap-east-1. Greenpixie's Region Scorecard grades cloud regions on carbon, water stress and price, so teams can shortlist regions before they deploy.
What are the benefits of green cloud computing?
Lower cloud bills, a smaller footprint and better data for sustainability reporting and for tracking sustainability goals. Most changes that cut energy use also cut cost, and resource-level footprint data gives engineering teams a second reason to act on optimisation work that cost alone has not moved.
Sources
- IEA, Data centre electricity use surged in 2025, April 2026
- Masanet et al., Recalibrating global data center energy-use estimates, Science, February 2020
- Uptime Institute, Global Data Center Survey 2026
- AWS, Sustainability
- Microsoft, Datacenter efficiency
- Google, Data center efficiency
- Amazon, 2025 Sustainability Report summary
- Google, 2026 Environmental Report
- GHG Protocol, Scope 2 Guidance
- Google, Our data centers now work harder when the sun shines and wind blows, April 2020
- Green Software Foundation, Carbon Aware SDK
- Greenpixie, GreenOps and the Mastercard results
- James Hall, GreenOps through FinOps, techUK, September 2025

