As the majority of businesses now rely on cloud and digital infrastructure, our expert Chiara Petrillo, uncovers why calculating greenhouse gas emissions from these services is a complex exercise.
As organisations accelerate digital transformation, cloud computing has become fundamental to business operations. From enterprise software and data storage to artificial intelligence (AI) applications, cloud-based services now underpin much of the global economy.
Yet as reliance on digital infrastructure grows, so too does its environmental footprint. Electricity demand from data centres, artificial intelligence (AI) and cryptocurrencies reached approximately 460 TWh in 2022, almost 2% of global demand, and could exceed 1,000 TWh by 2026. To put this into perspective, a single hyperscale1 data centre consumes as much electricity as a medium-sized town.
For organisations pursuing climate targets an important question is emerging: how accurately can businesses calculate the greenhouse gas (GHG) emissions associated with the cloud services and digital infrastructure they depend on?
Moving to the cloud does not eliminate emissions — it changes how they are accounted for
A common assumption is that migrating from on premises infrastructure to cloud services automatically reduces emissions. While cloud computing can improve efficiency in some circumstances, the reality is more nuanced.
When organisations move data storage, software platforms, or computing workloads to a third-party cloud provider, emissions associated with electricity use shift from Scope 2 to Scope 3 under the Greenhouse Gas Protocol.
The physical activity remains. Servers still need electricity, cooling systems still operate continuously, and digital services still require energy-intensive infrastructure to function. What changes is where those emissions appear in an organisation’s GHG inventory.
This distinction matters. Companies remain accountable for emissions associated with cloud-based services, but they often have limited control over (and limited visibility into) how the underlying electricity running the data centres is generated. As a result, a transfer of infrastructure can create greater distance between organisations and the emissions they are expected to understand and report.
Why data centre emissions matter more than ever
Data centres currently account for around 1–1.5% of global electricity consumption and demand is increasing rapidly. This level of demand is comparable to the electricity consumption of large, industrialised economies.
AI workloads are a significant driver of growth. AI workloads require substantially greater computing power than more traditional applications, increasing both electricity demand for processing capacity and cooling infrastructure. In large-scale facilities, cooling can account for up to around half of total energy use.
As demand grows, electricity use is becoming more concentrated in large-scale facilities, increasing pressure on local electricity systems and, in some locations, placing additional demands on local water resources used for cooling.5
The GHG footprint of cloud computing depends on geography
Cloud services are often perceived as virtual, but in reality they run on highly physical infrastructure. Every cloud service ultimately runs through a specific data centre connected to a specific electricity grid. This means the carbon intensity of cloud computing is heavily influenced by geography.
A workload hosted in a region where electricity generation relies heavily on fossil fuels, such as Northern Virginia (USA) for example, may be associated with significantly higher emissions than an identical workload running in a region supplied primarily by renewable energy, hydroelectricity or nuclear power, Sweden or Finland for example 3. This is one reason why major cloud providers, including Google, Microsoft and Meta, have invested in countries such as Sweden and Finland, where electricity systems tend to have comparatively lower carbon intensity.
As a result, two organisations using the same cloud service could be responsible for materially different emissions depending on where electricity is generated. This creates an important challenge for emissions reporting. Emissions are typically disclosed at the service level, but their real-world environmental impact is determined by the electricity systems supporting the infrastructure behind that service.
The growing transparency gap in cloud emissions reporting
Understanding cloud-related emissions is not simply a question of measuring electricity consumption. It is also a question of how emissions are reported.
Most cloud-related emissions reporting follows market-based accounting approaches aligned with the Greenhouse Gas Protocol. These approaches may incorporate contractual instruments such as renewable energy certificates and power purchase agreements. While these mechanisms can play an important role in supporting renewable energy development, they do not necessarily reflect the carbon intensity of the electricity physically consumed at a specific location and time.
Location-based reporting, by contrast, seeks to reflect the actual emissions associated with the grid supplying electricity to a facility. The distinction is important because companies are increasingly expected to report emissions associated with cloud services, yet often lack access to the underlying data needed to understand the physical carbon impact of those services.
This creates a transparency gap between reported emissions and real-world emissions. Why does this matter now?
Several trends are converging to make digital infrastructure emissions a strategic business issue.
Demand for cloud computing and AI services continues to accelerate. At the same time, data centres are becoming increasingly important drivers of electricity demand growth and water use in major economies.
Alongside this, sustainability reporting requirements are becoming more comprehensive. Under the Corporate Sustainability Reporting Directive (CSRD), organisations are expected to provide greater transparency across Scope 1, Scope 2 and Scope 3 emissions, including emissions associated with outsourced digital services.
As scrutiny of corporate climate claims grows, businesses need confidence that reported emissions provide a meaningful representation of environmental impact.
Improving understanding of digital infrastructure emissions may require organisations to go beyond standard disclosures by engaging cloud providers on the location of workloads, seeking greater transparency on location-based emissions, and understanding the role that renewable energy certificates and other contractual instruments play within reported figures.
As the digital economy continues to expand, emissions associated with cloud services, AI infrastructure and data centres are likely to become increasingly significant within Scope 3 inventories.
Businesses that gain a clearer understanding of these emissions today will be better positioned to develop robust climate strategies, meet evolving reporting requirements, manage climate-related risks, andmake more informed decisions about the digital infrastructure that supports their operations.