Incorporating sustainability criteria and metrics into FinOps activities to ensure environmental efficiency is balanced with financial value, and optimization decisions are aligned with organizational goals.
Sustainability defines how the organization will make decisions about using technology in ways that consider both its impact on the environment and the organization’s broader sustainability goals. Sustainability considerations allow engineers and product personas to balance environmental value alongside financial benefits when architecting, optimizing, and deploying workloads across all technology domains, including cloud, data centers, data cloud platforms, AI, and other infrastructure.
Many organizations have a sustainability program that looks broadly at environmental impact beyond technology. The proportion of environmental impact attributable to technology use will vary greatly by organization depending on all of its sources of carbon emissions. Regulation requiring more regular reporting of environmental impacts, including direct and indirect carbon emissions, is being enacted in many areas of the world. As a result, visibility into carbon use across all technology domains is becoming increasingly important for cost allocation, reporting, forecasting, and other critical IT functions.
FinOps teams should be integrating with organizational sustainability programs by leveraging the Intersecting Disciplines Capability. By incorporating sustainability information into the activities performed under the Understand Usage & Cost and Quantify Business Value Domains, along with activities from the Sustainability Capability, practitioners should be working to identify opportunities for optimization of technology cost and usage to support the organization’s sustainability goals.
Activities done in this Capability will generate potential opportunities to reduce the carbon footprint of technology use across all domains, including cloud, data centers, data platforms, and AI. Recommendations for optimization opportunities can be evaluated by the FinOps team and other Personas to select the options over time that produce the best value to the organization, whether financial, environmental, or operational.
Usage Optimization recommendations correlate closely with lower carbon emissions by only using what is needed, and only when it is needed. For example, right-size workloads to match actual demand, rationalize unused SaaS licenses, or refresh on-premises equipment on an efficient lifecycle schedule.
However, sustainability efforts can run counter to cost savings in Rate Optimization efforts, where reserved discounts on certain resources might discourage optimizing them or scaling them down. Some architectural or operational decisions may conflict with sustainability goals, for example operational policies to run blade servers at only 30% maximum CPU utilization to support availability, or the use of nearby (but less energy efficient) data center co-locations for latency reasons. And some technology resources or services in certain locations may be less expensive but more carbon intensive (or vice versa), leading to tradeoffs when deciding where to deploy workloads.
In order to achieve outcomes related to Sustainability, organizations need to consider embodied carbon, which captures the full lifecycle of their technology resources from the sourcing of materials and energy used to build and operate infrastructure, to the disposal of outdated equipment. Embodied carbon considerations extend across all FinOps Scopes, including the manufacturing and disposal of on-premises servers and networking equipment, hardware refresh cycles for end-user computing devices, and the upstream emissions associated with third-party SaaS platforms.
Organizations that take a comprehensive view of embodied carbon across all of these categories will be better positioned to accurately measure and report on their full technology-related carbon footprint, and to identify the highest impact opportunities for reduction.
AI and machine learning workloads deserve specific attention within any sustainability strategy. Training and inference workloads running on GPU clusters are materially more energy-intensive than general-purpose compute, and their carbon profile varies significantly based on the energy mix of the region where they run.
As AI spend becomes a standard part of the technology portfolio — and as organizations scale from experimentation to production — the sustainability implications of AI infrastructure choices become a meaningful input to both workload placement decisions and vendor selection. FinOps teams supporting AI cost management should extend that same lens to carbon efficiency, and consider treating AI as a distinct category within the organization’s sustainability accounting.
As with all topics in FinOps, collaboration among teams is crucial to determine the top priorities for the organization and allowing those priorities to determine how to balance tradeoffs. Sustainability must be considered in the context of the needs of the business. The iron triangle balancing of cost, speed, and quality must weigh the cost and benefits of sustainability activities.
While the Sustainability Capability is part of the Optimize Usage and Cost Domain, sustainability data will be used in many Capabilities across other Domains as well. For example, sustainability data may be incorporated as part of the Data Ingestion Capability, the Allocation Capability, included in Reporting & Analytics activities, used in metrics to support Unit Economics, or used in KPI & Benchmarking activities by engineering and other practitioner teams to track and improve emissions performance.
Sustainability considerations will include items such as the energy used to power and cool the servers and infrastructure that underpin all technology categories, as well as the efficiency of architectures, and placement of workloads. The degree to which organizations can optimize their carbon footprint will vary significantly by technology category.
Cloud environments offer dynamic, on-demand optimization opportunities that can be acted upon relatively quickly, while on-premises data centers and colocation facilities are constrained by fixed hardware investments and longer refresh cycles that require a more strategic and planned approach to carbon reduction. SaaS platforms and third-party services present a different set of considerations, where the organization’s ability to directly influence carbon outcomes is limited and largely dependent on the sustainability commitments and reporting transparency of the vendor.
End-user computing introduces additional complexity through device procurement, lifecycle management, and disposal practices that contribute meaningfully to the organization’s overall technology carbon footprint. Recognizing these differences is essential for developing a realistic and effective Sustainability strategy, and for setting appropriate expectations about where and how quickly carbon reductions can be achieved.
Sustainability data available from data providers, technology vendors, and billing sources will of course vary in its scope, granularity, and quality. FinOps teams will need to collaborate with Sustainability Personas to determine the necessary granularity and how best to adjust or normalize data from various sources to align with corporate mandates. Sustainability data quality concerns may continue for some time, and organizations should provide input to their technology providers and vendors about the data they require. However, the primary goal of this Capability is to use whatever data is available, even when it is incomplete, to make recommendations to the organization regarding sustainable technology use. Assumptions, estimation methodologies, and data quality limitations should be clearly documented to ensure transparency and reduce the risk of misinterpretation or regulatory exposure. Architectural, usage efficiency, and rate optimization decisions will continue to be made without this input if all efforts are centered on data quality rather than on using data to make directionally-correct, if imperfect, recommendations about sustainable use across all technology domains.
There is also a meaningful relationship between optimization activity and sustainability investment. Organizations that systematically reduce waste create both financial headroom and carbon reductions simultaneously. Some organizations are beginning to treat this dynamic explicitly, using efficiency gains to fund sustainability improvements such as workload migration to lower-carbon regions or infrastructure refresh toward more energy-efficient hardware. Making this relationship visible in planning and reporting reinforces sustainability as a natural output of good FinOps practice, not a competing priority.
As someone in the FinOps team role, I will…
As someone in a Product role, I will…
As someone in a Finance role, I will…
As someone in a Procurement role, I will…
As someone in an Engineering role, I will…
As someone in a Leadership role, I will…
As someone in an Allied Persona role, I will…
Key terms that are likely to be used in sustainability reports:
| Term | Description |
| Metric tons of carbon dioxide equivalents (mtCO2e) | Each measurement of greenhouse gas can be converted to metric tons of carbon dioxide equivalents by using that greenhouse gas’s global warming potential (GWP) factor. |
| kg of carbon dioxide equivalents (kgCO2e) | Each measurement of greenhouse gas can be converted to kilograms (kg) of carbon dioxide equivalents by using that greenhouse gas’s global warming potential (GWP) factor. |
| Liters of H2O consumed | Measurement for water consumption (commonly used in data centers) |
| MWh of electricity | Total megawatt hours (MWh) used from electricity |
| kWh of electricity | Total kilowatt hours (kWh) used from electricity |
| m3 of water | Typically, total water consumed in cubic meters |
| ESG | Environmental, Sustainability & Governance – sometimes an umbrella term for where Sustainability efforts are run in organizations. |
| CO2e | Carbon Dioxide Equivalent, or Carbon Equivalent, sometimes used as a unified way of expressing the environmental impact of an activity taking into account all of the various emissions or resource uses and expressing them in an equivalent of carbon dioxide for easier reporting with a unified measurement |
| SDG | Sustainable Development Goals, a set of evolving sustainability goals established and reported upon by various UN and governmental organizations |
| GHG | Greenhouse Gasses, all of the various gasses which affect the environment as emissions, including carbon dioxide, methane, and other pollutants |
This KPI measures the amount of compliance for cloud sustainability tagging.
This KPI measures the amount of compliance for cloud sustainability tagging, and requires an established organizational tagging policy. The sophistication of determining what acceptable tag criteria are, and the stringency of the KPI should evolve with the organization’s FinOps maturity.
(Total emissions Associated with Tagging Policy Compliant CSP Cloud Resources During a Period of Time / Total CSP Cloud Emissions During a Period of Time) x 100
Data Sources:
This metric compares present forecasted vs. actual cloud emissions over a specific period (e.g., day, month, quarter).
These 2 KPIs measure carbon inefficiency in workload utilization. It quantifies the difference in carbon emissions between a fully optimized resource allocation and the current allocation. This approach aligns directly with the FinOps principle of optimizing workloads to maximize resource utilization by selecting the most suitable resource size and type. The optimized allocation considers factors such as architecture type (e.g., x86 vs. AMD), as well as specific resource attributes like type and size.
Carbon Waste Formula:
Carbon Efficiency Formula:
Data Sources:
CSP Cloud Carbon Budget Burn RateCarbon emissions per unit of cloud service SKU measure the amount of carbon dioxide equivalent (CO2e) emitted to deliver a specific unit of cloud services, such as compute, storage, and/or networking.
Carbon emissions per unit of cloud service SKU measure the amount of carbon dioxide equivalent (CO2e) emitted to deliver a specific unit of cloud services, such as compute, storage, and/or networking. This metric allows businesses and consumers to track the environmental impact of cloud usage by resource and make decisions to reduce their carbon footprint by choosing more efficient or renewable-energy-powered solutions.
Carbon per unit (kgCO2e) = Total Carbon Emissions (kgCO2e) / Total Resource Usage (SKU)
Data Sources:
Additional Guidance:
Measure of carbon emissions (CO2e) per unit of cloud spend. Ideally, this measure would be drillable by scope and service categories, such as compute, storage, or networking.
Measure of carbon emissions (CO2e) per unit of cloud spend. Ideally, this measure would be drillable by scope and service categories, such as compute, storage, or networking. This metric helps organizations understand the environmental impact of cloud usage and make decisions to reduce their carbon footprint by choosing more efficient or renewable-energy-powered cloud services.
CO2e/unit of currency = Effective Cost / CO2e
Data Sources:
Forecast Accuracy Rate for Carbon Emissions
This metric compares forecasted vs. actual cloud emissions over a specific period (e.g., day, month, quarter). While percentage variance is the primary metric, carbon differences can also be informative. Each organization defines its acceptable variance.
((Forecasted Public Cloud emissions – Actual Public Cloud emissions) / Forecasted Public Cloud emissions)
Data Sources: